Show More
@@ -1,795 +1,805 b'' | |||
|
1 | 1 | /* |
|
2 | 2 | parsers.c - efficient content parsing |
|
3 | 3 | |
|
4 | 4 | Copyright 2008 Olivia Mackall <olivia@selenic.com> and others |
|
5 | 5 | |
|
6 | 6 | This software may be used and distributed according to the terms of |
|
7 | 7 | the GNU General Public License, incorporated herein by reference. |
|
8 | 8 | */ |
|
9 | 9 | |
|
10 | 10 | #define PY_SSIZE_T_CLEAN |
|
11 | 11 | #include <Python.h> |
|
12 | 12 | #include <ctype.h> |
|
13 | 13 | #include <stddef.h> |
|
14 | 14 | #include <string.h> |
|
15 | 15 | |
|
16 | 16 | #include "bitmanipulation.h" |
|
17 | 17 | #include "charencode.h" |
|
18 | 18 | #include "util.h" |
|
19 | 19 | |
|
20 | 20 | #ifdef IS_PY3K |
|
21 | 21 | /* The mapping of Python types is meant to be temporary to get Python |
|
22 | 22 | * 3 to compile. We should remove this once Python 3 support is fully |
|
23 | 23 | * supported and proper types are used in the extensions themselves. */ |
|
24 | 24 | #define PyInt_Check PyLong_Check |
|
25 | 25 | #define PyInt_FromLong PyLong_FromLong |
|
26 | 26 | #define PyInt_FromSsize_t PyLong_FromSsize_t |
|
27 | 27 | #define PyInt_AsLong PyLong_AsLong |
|
28 | 28 | #endif |
|
29 | 29 | |
|
30 | 30 | static const char *const versionerrortext = "Python minor version mismatch"; |
|
31 | 31 | |
|
32 | 32 | static PyObject *dict_new_presized(PyObject *self, PyObject *args) |
|
33 | 33 | { |
|
34 | 34 | Py_ssize_t expected_size; |
|
35 | 35 | |
|
36 | 36 | if (!PyArg_ParseTuple(args, "n:make_presized_dict", &expected_size)) { |
|
37 | 37 | return NULL; |
|
38 | 38 | } |
|
39 | 39 | |
|
40 | 40 | return _dict_new_presized(expected_size); |
|
41 | 41 | } |
|
42 | 42 | |
|
43 | 43 | static inline dirstateTupleObject *make_dirstate_tuple(char state, int mode, |
|
44 | 44 | int size, int mtime) |
|
45 | 45 | { |
|
46 | 46 | dirstateTupleObject *t = |
|
47 | 47 | PyObject_New(dirstateTupleObject, &dirstateTupleType); |
|
48 | 48 | if (!t) { |
|
49 | 49 | return NULL; |
|
50 | 50 | } |
|
51 | 51 | t->state = state; |
|
52 | 52 | t->mode = mode; |
|
53 | 53 | t->size = size; |
|
54 | 54 | t->mtime = mtime; |
|
55 | 55 | return t; |
|
56 | 56 | } |
|
57 | 57 | |
|
58 | 58 | static PyObject *dirstate_tuple_new(PyTypeObject *subtype, PyObject *args, |
|
59 | 59 | PyObject *kwds) |
|
60 | 60 | { |
|
61 | 61 | /* We do all the initialization here and not a tp_init function because |
|
62 | 62 | * dirstate_tuple is immutable. */ |
|
63 | 63 | dirstateTupleObject *t; |
|
64 | 64 | char state; |
|
65 | 65 | int size, mode, mtime; |
|
66 | 66 | if (!PyArg_ParseTuple(args, "ciii", &state, &mode, &size, &mtime)) { |
|
67 | 67 | return NULL; |
|
68 | 68 | } |
|
69 | 69 | |
|
70 | 70 | t = (dirstateTupleObject *)subtype->tp_alloc(subtype, 1); |
|
71 | 71 | if (!t) { |
|
72 | 72 | return NULL; |
|
73 | 73 | } |
|
74 | 74 | t->state = state; |
|
75 | 75 | t->mode = mode; |
|
76 | 76 | t->size = size; |
|
77 | 77 | t->mtime = mtime; |
|
78 | 78 | |
|
79 | 79 | return (PyObject *)t; |
|
80 | 80 | } |
|
81 | 81 | |
|
82 | 82 | static void dirstate_tuple_dealloc(PyObject *o) |
|
83 | 83 | { |
|
84 | 84 | PyObject_Del(o); |
|
85 | 85 | } |
|
86 | 86 | |
|
87 | 87 | static Py_ssize_t dirstate_tuple_length(PyObject *o) |
|
88 | 88 | { |
|
89 | 89 | return 4; |
|
90 | 90 | } |
|
91 | 91 | |
|
92 | 92 | static PyObject *dirstate_tuple_item(PyObject *o, Py_ssize_t i) |
|
93 | 93 | { |
|
94 | 94 | dirstateTupleObject *t = (dirstateTupleObject *)o; |
|
95 | 95 | switch (i) { |
|
96 | 96 | case 0: |
|
97 | 97 | return PyBytes_FromStringAndSize(&t->state, 1); |
|
98 | 98 | case 1: |
|
99 | 99 | return PyInt_FromLong(t->mode); |
|
100 | 100 | case 2: |
|
101 | 101 | return PyInt_FromLong(t->size); |
|
102 | 102 | case 3: |
|
103 | 103 | return PyInt_FromLong(t->mtime); |
|
104 | 104 | default: |
|
105 | 105 | PyErr_SetString(PyExc_IndexError, "index out of range"); |
|
106 | 106 | return NULL; |
|
107 | 107 | } |
|
108 | 108 | } |
|
109 | 109 | |
|
110 | 110 | static PySequenceMethods dirstate_tuple_sq = { |
|
111 | 111 | dirstate_tuple_length, /* sq_length */ |
|
112 | 112 | 0, /* sq_concat */ |
|
113 | 113 | 0, /* sq_repeat */ |
|
114 | 114 | dirstate_tuple_item, /* sq_item */ |
|
115 | 115 | 0, /* sq_ass_item */ |
|
116 | 116 | 0, /* sq_contains */ |
|
117 | 117 | 0, /* sq_inplace_concat */ |
|
118 | 118 | 0 /* sq_inplace_repeat */ |
|
119 | 119 | }; |
|
120 | 120 | |
|
121 | 121 | static PyObject *dirstatetuple_v1_state(dirstateTupleObject *self) |
|
122 | 122 | { |
|
123 | 123 | return PyBytes_FromStringAndSize(&self->state, 1); |
|
124 | 124 | }; |
|
125 | 125 | |
|
126 | 126 | static PyObject *dirstatetuple_v1_mode(dirstateTupleObject *self) |
|
127 | 127 | { |
|
128 | 128 | return PyInt_FromLong(self->mode); |
|
129 | 129 | }; |
|
130 | 130 | |
|
131 | 131 | static PyObject *dirstatetuple_v1_size(dirstateTupleObject *self) |
|
132 | 132 | { |
|
133 | 133 | return PyInt_FromLong(self->size); |
|
134 | 134 | }; |
|
135 | 135 | |
|
136 | 136 | static PyObject *dirstatetuple_v1_mtime(dirstateTupleObject *self) |
|
137 | 137 | { |
|
138 | 138 | return PyInt_FromLong(self->mtime); |
|
139 | 139 | }; |
|
140 | 140 | |
|
141 | 141 | static PyMethodDef dirstatetuple_methods[] = { |
|
142 | 142 | {"v1_state", (PyCFunction)dirstatetuple_v1_state, METH_NOARGS, |
|
143 | 143 | "return a \"state\" suitable for v1 serialization"}, |
|
144 | 144 | {"v1_mode", (PyCFunction)dirstatetuple_v1_mode, METH_NOARGS, |
|
145 | 145 | "return a \"mode\" suitable for v1 serialization"}, |
|
146 | 146 | {"v1_size", (PyCFunction)dirstatetuple_v1_size, METH_NOARGS, |
|
147 | 147 | "return a \"size\" suitable for v1 serialization"}, |
|
148 | 148 | {"v1_mtime", (PyCFunction)dirstatetuple_v1_mtime, METH_NOARGS, |
|
149 | 149 | "return a \"mtime\" suitable for v1 serialization"}, |
|
150 | 150 | {NULL} /* Sentinel */ |
|
151 | 151 | }; |
|
152 | 152 | |
|
153 | static PyObject *dirstatetuple_get_state(dirstateTupleObject *self) | |
|
154 | { | |
|
155 | return PyBytes_FromStringAndSize(&self->state, 1); | |
|
156 | }; | |
|
157 | ||
|
158 | static PyGetSetDef dirstatetuple_getset[] = { | |
|
159 | {"state", (getter)dirstatetuple_get_state, NULL, "state", NULL}, | |
|
160 | {NULL} /* Sentinel */ | |
|
161 | }; | |
|
162 | ||
|
153 | 163 | PyTypeObject dirstateTupleType = { |
|
154 | 164 | PyVarObject_HEAD_INIT(NULL, 0) /* header */ |
|
155 | 165 | "dirstate_tuple", /* tp_name */ |
|
156 | 166 | sizeof(dirstateTupleObject), /* tp_basicsize */ |
|
157 | 167 | 0, /* tp_itemsize */ |
|
158 | 168 | (destructor)dirstate_tuple_dealloc, /* tp_dealloc */ |
|
159 | 169 | 0, /* tp_print */ |
|
160 | 170 | 0, /* tp_getattr */ |
|
161 | 171 | 0, /* tp_setattr */ |
|
162 | 172 | 0, /* tp_compare */ |
|
163 | 173 | 0, /* tp_repr */ |
|
164 | 174 | 0, /* tp_as_number */ |
|
165 | 175 | &dirstate_tuple_sq, /* tp_as_sequence */ |
|
166 | 176 | 0, /* tp_as_mapping */ |
|
167 | 177 | 0, /* tp_hash */ |
|
168 | 178 | 0, /* tp_call */ |
|
169 | 179 | 0, /* tp_str */ |
|
170 | 180 | 0, /* tp_getattro */ |
|
171 | 181 | 0, /* tp_setattro */ |
|
172 | 182 | 0, /* tp_as_buffer */ |
|
173 | 183 | Py_TPFLAGS_DEFAULT, /* tp_flags */ |
|
174 | 184 | "dirstate tuple", /* tp_doc */ |
|
175 | 185 | 0, /* tp_traverse */ |
|
176 | 186 | 0, /* tp_clear */ |
|
177 | 187 | 0, /* tp_richcompare */ |
|
178 | 188 | 0, /* tp_weaklistoffset */ |
|
179 | 189 | 0, /* tp_iter */ |
|
180 | 190 | 0, /* tp_iternext */ |
|
181 | 191 | dirstatetuple_methods, /* tp_methods */ |
|
182 | 192 | 0, /* tp_members */ |
|
183 |
|
|
|
193 | dirstatetuple_getset, /* tp_getset */ | |
|
184 | 194 | 0, /* tp_base */ |
|
185 | 195 | 0, /* tp_dict */ |
|
186 | 196 | 0, /* tp_descr_get */ |
|
187 | 197 | 0, /* tp_descr_set */ |
|
188 | 198 | 0, /* tp_dictoffset */ |
|
189 | 199 | 0, /* tp_init */ |
|
190 | 200 | 0, /* tp_alloc */ |
|
191 | 201 | dirstate_tuple_new, /* tp_new */ |
|
192 | 202 | }; |
|
193 | 203 | |
|
194 | 204 | static PyObject *parse_dirstate(PyObject *self, PyObject *args) |
|
195 | 205 | { |
|
196 | 206 | PyObject *dmap, *cmap, *parents = NULL, *ret = NULL; |
|
197 | 207 | PyObject *fname = NULL, *cname = NULL, *entry = NULL; |
|
198 | 208 | char state, *cur, *str, *cpos; |
|
199 | 209 | int mode, size, mtime; |
|
200 | 210 | unsigned int flen, pos = 40; |
|
201 | 211 | Py_ssize_t len = 40; |
|
202 | 212 | Py_ssize_t readlen; |
|
203 | 213 | |
|
204 | 214 | if (!PyArg_ParseTuple( |
|
205 | 215 | args, PY23("O!O!s#:parse_dirstate", "O!O!y#:parse_dirstate"), |
|
206 | 216 | &PyDict_Type, &dmap, &PyDict_Type, &cmap, &str, &readlen)) { |
|
207 | 217 | goto quit; |
|
208 | 218 | } |
|
209 | 219 | |
|
210 | 220 | len = readlen; |
|
211 | 221 | |
|
212 | 222 | /* read parents */ |
|
213 | 223 | if (len < 40) { |
|
214 | 224 | PyErr_SetString(PyExc_ValueError, |
|
215 | 225 | "too little data for parents"); |
|
216 | 226 | goto quit; |
|
217 | 227 | } |
|
218 | 228 | |
|
219 | 229 | parents = Py_BuildValue(PY23("s#s#", "y#y#"), str, (Py_ssize_t)20, |
|
220 | 230 | str + 20, (Py_ssize_t)20); |
|
221 | 231 | if (!parents) { |
|
222 | 232 | goto quit; |
|
223 | 233 | } |
|
224 | 234 | |
|
225 | 235 | /* read filenames */ |
|
226 | 236 | while (pos >= 40 && pos < len) { |
|
227 | 237 | if (pos + 17 > len) { |
|
228 | 238 | PyErr_SetString(PyExc_ValueError, |
|
229 | 239 | "overflow in dirstate"); |
|
230 | 240 | goto quit; |
|
231 | 241 | } |
|
232 | 242 | cur = str + pos; |
|
233 | 243 | /* unpack header */ |
|
234 | 244 | state = *cur; |
|
235 | 245 | mode = getbe32(cur + 1); |
|
236 | 246 | size = getbe32(cur + 5); |
|
237 | 247 | mtime = getbe32(cur + 9); |
|
238 | 248 | flen = getbe32(cur + 13); |
|
239 | 249 | pos += 17; |
|
240 | 250 | cur += 17; |
|
241 | 251 | if (flen > len - pos) { |
|
242 | 252 | PyErr_SetString(PyExc_ValueError, |
|
243 | 253 | "overflow in dirstate"); |
|
244 | 254 | goto quit; |
|
245 | 255 | } |
|
246 | 256 | |
|
247 | 257 | entry = |
|
248 | 258 | (PyObject *)make_dirstate_tuple(state, mode, size, mtime); |
|
249 | 259 | cpos = memchr(cur, 0, flen); |
|
250 | 260 | if (cpos) { |
|
251 | 261 | fname = PyBytes_FromStringAndSize(cur, cpos - cur); |
|
252 | 262 | cname = PyBytes_FromStringAndSize( |
|
253 | 263 | cpos + 1, flen - (cpos - cur) - 1); |
|
254 | 264 | if (!fname || !cname || |
|
255 | 265 | PyDict_SetItem(cmap, fname, cname) == -1 || |
|
256 | 266 | PyDict_SetItem(dmap, fname, entry) == -1) { |
|
257 | 267 | goto quit; |
|
258 | 268 | } |
|
259 | 269 | Py_DECREF(cname); |
|
260 | 270 | } else { |
|
261 | 271 | fname = PyBytes_FromStringAndSize(cur, flen); |
|
262 | 272 | if (!fname || |
|
263 | 273 | PyDict_SetItem(dmap, fname, entry) == -1) { |
|
264 | 274 | goto quit; |
|
265 | 275 | } |
|
266 | 276 | } |
|
267 | 277 | Py_DECREF(fname); |
|
268 | 278 | Py_DECREF(entry); |
|
269 | 279 | fname = cname = entry = NULL; |
|
270 | 280 | pos += flen; |
|
271 | 281 | } |
|
272 | 282 | |
|
273 | 283 | ret = parents; |
|
274 | 284 | Py_INCREF(ret); |
|
275 | 285 | quit: |
|
276 | 286 | Py_XDECREF(fname); |
|
277 | 287 | Py_XDECREF(cname); |
|
278 | 288 | Py_XDECREF(entry); |
|
279 | 289 | Py_XDECREF(parents); |
|
280 | 290 | return ret; |
|
281 | 291 | } |
|
282 | 292 | |
|
283 | 293 | /* |
|
284 | 294 | * Build a set of non-normal and other parent entries from the dirstate dmap |
|
285 | 295 | */ |
|
286 | 296 | static PyObject *nonnormalotherparententries(PyObject *self, PyObject *args) |
|
287 | 297 | { |
|
288 | 298 | PyObject *dmap, *fname, *v; |
|
289 | 299 | PyObject *nonnset = NULL, *otherpset = NULL, *result = NULL; |
|
290 | 300 | Py_ssize_t pos; |
|
291 | 301 | |
|
292 | 302 | if (!PyArg_ParseTuple(args, "O!:nonnormalentries", &PyDict_Type, |
|
293 | 303 | &dmap)) { |
|
294 | 304 | goto bail; |
|
295 | 305 | } |
|
296 | 306 | |
|
297 | 307 | nonnset = PySet_New(NULL); |
|
298 | 308 | if (nonnset == NULL) { |
|
299 | 309 | goto bail; |
|
300 | 310 | } |
|
301 | 311 | |
|
302 | 312 | otherpset = PySet_New(NULL); |
|
303 | 313 | if (otherpset == NULL) { |
|
304 | 314 | goto bail; |
|
305 | 315 | } |
|
306 | 316 | |
|
307 | 317 | pos = 0; |
|
308 | 318 | while (PyDict_Next(dmap, &pos, &fname, &v)) { |
|
309 | 319 | dirstateTupleObject *t; |
|
310 | 320 | if (!dirstate_tuple_check(v)) { |
|
311 | 321 | PyErr_SetString(PyExc_TypeError, |
|
312 | 322 | "expected a dirstate tuple"); |
|
313 | 323 | goto bail; |
|
314 | 324 | } |
|
315 | 325 | t = (dirstateTupleObject *)v; |
|
316 | 326 | |
|
317 | 327 | if (t->state == 'n' && t->size == -2) { |
|
318 | 328 | if (PySet_Add(otherpset, fname) == -1) { |
|
319 | 329 | goto bail; |
|
320 | 330 | } |
|
321 | 331 | } |
|
322 | 332 | |
|
323 | 333 | if (t->state == 'n' && t->mtime != -1) { |
|
324 | 334 | continue; |
|
325 | 335 | } |
|
326 | 336 | if (PySet_Add(nonnset, fname) == -1) { |
|
327 | 337 | goto bail; |
|
328 | 338 | } |
|
329 | 339 | } |
|
330 | 340 | |
|
331 | 341 | result = Py_BuildValue("(OO)", nonnset, otherpset); |
|
332 | 342 | if (result == NULL) { |
|
333 | 343 | goto bail; |
|
334 | 344 | } |
|
335 | 345 | Py_DECREF(nonnset); |
|
336 | 346 | Py_DECREF(otherpset); |
|
337 | 347 | return result; |
|
338 | 348 | bail: |
|
339 | 349 | Py_XDECREF(nonnset); |
|
340 | 350 | Py_XDECREF(otherpset); |
|
341 | 351 | Py_XDECREF(result); |
|
342 | 352 | return NULL; |
|
343 | 353 | } |
|
344 | 354 | |
|
345 | 355 | /* |
|
346 | 356 | * Efficiently pack a dirstate object into its on-disk format. |
|
347 | 357 | */ |
|
348 | 358 | static PyObject *pack_dirstate(PyObject *self, PyObject *args) |
|
349 | 359 | { |
|
350 | 360 | PyObject *packobj = NULL; |
|
351 | 361 | PyObject *map, *copymap, *pl, *mtime_unset = NULL; |
|
352 | 362 | Py_ssize_t nbytes, pos, l; |
|
353 | 363 | PyObject *k, *v = NULL, *pn; |
|
354 | 364 | char *p, *s; |
|
355 | 365 | int now; |
|
356 | 366 | |
|
357 | 367 | if (!PyArg_ParseTuple(args, "O!O!O!i:pack_dirstate", &PyDict_Type, &map, |
|
358 | 368 | &PyDict_Type, ©map, &PyTuple_Type, &pl, |
|
359 | 369 | &now)) { |
|
360 | 370 | return NULL; |
|
361 | 371 | } |
|
362 | 372 | |
|
363 | 373 | if (PyTuple_Size(pl) != 2) { |
|
364 | 374 | PyErr_SetString(PyExc_TypeError, "expected 2-element tuple"); |
|
365 | 375 | return NULL; |
|
366 | 376 | } |
|
367 | 377 | |
|
368 | 378 | /* Figure out how much we need to allocate. */ |
|
369 | 379 | for (nbytes = 40, pos = 0; PyDict_Next(map, &pos, &k, &v);) { |
|
370 | 380 | PyObject *c; |
|
371 | 381 | if (!PyBytes_Check(k)) { |
|
372 | 382 | PyErr_SetString(PyExc_TypeError, "expected string key"); |
|
373 | 383 | goto bail; |
|
374 | 384 | } |
|
375 | 385 | nbytes += PyBytes_GET_SIZE(k) + 17; |
|
376 | 386 | c = PyDict_GetItem(copymap, k); |
|
377 | 387 | if (c) { |
|
378 | 388 | if (!PyBytes_Check(c)) { |
|
379 | 389 | PyErr_SetString(PyExc_TypeError, |
|
380 | 390 | "expected string key"); |
|
381 | 391 | goto bail; |
|
382 | 392 | } |
|
383 | 393 | nbytes += PyBytes_GET_SIZE(c) + 1; |
|
384 | 394 | } |
|
385 | 395 | } |
|
386 | 396 | |
|
387 | 397 | packobj = PyBytes_FromStringAndSize(NULL, nbytes); |
|
388 | 398 | if (packobj == NULL) { |
|
389 | 399 | goto bail; |
|
390 | 400 | } |
|
391 | 401 | |
|
392 | 402 | p = PyBytes_AS_STRING(packobj); |
|
393 | 403 | |
|
394 | 404 | pn = PyTuple_GET_ITEM(pl, 0); |
|
395 | 405 | if (PyBytes_AsStringAndSize(pn, &s, &l) == -1 || l != 20) { |
|
396 | 406 | PyErr_SetString(PyExc_TypeError, "expected a 20-byte hash"); |
|
397 | 407 | goto bail; |
|
398 | 408 | } |
|
399 | 409 | memcpy(p, s, l); |
|
400 | 410 | p += 20; |
|
401 | 411 | pn = PyTuple_GET_ITEM(pl, 1); |
|
402 | 412 | if (PyBytes_AsStringAndSize(pn, &s, &l) == -1 || l != 20) { |
|
403 | 413 | PyErr_SetString(PyExc_TypeError, "expected a 20-byte hash"); |
|
404 | 414 | goto bail; |
|
405 | 415 | } |
|
406 | 416 | memcpy(p, s, l); |
|
407 | 417 | p += 20; |
|
408 | 418 | |
|
409 | 419 | for (pos = 0; PyDict_Next(map, &pos, &k, &v);) { |
|
410 | 420 | dirstateTupleObject *tuple; |
|
411 | 421 | char state; |
|
412 | 422 | int mode, size, mtime; |
|
413 | 423 | Py_ssize_t len, l; |
|
414 | 424 | PyObject *o; |
|
415 | 425 | char *t; |
|
416 | 426 | |
|
417 | 427 | if (!dirstate_tuple_check(v)) { |
|
418 | 428 | PyErr_SetString(PyExc_TypeError, |
|
419 | 429 | "expected a dirstate tuple"); |
|
420 | 430 | goto bail; |
|
421 | 431 | } |
|
422 | 432 | tuple = (dirstateTupleObject *)v; |
|
423 | 433 | |
|
424 | 434 | state = tuple->state; |
|
425 | 435 | mode = tuple->mode; |
|
426 | 436 | size = tuple->size; |
|
427 | 437 | mtime = tuple->mtime; |
|
428 | 438 | if (state == 'n' && mtime == now) { |
|
429 | 439 | /* See pure/parsers.py:pack_dirstate for why we do |
|
430 | 440 | * this. */ |
|
431 | 441 | mtime = -1; |
|
432 | 442 | mtime_unset = (PyObject *)make_dirstate_tuple( |
|
433 | 443 | state, mode, size, mtime); |
|
434 | 444 | if (!mtime_unset) { |
|
435 | 445 | goto bail; |
|
436 | 446 | } |
|
437 | 447 | if (PyDict_SetItem(map, k, mtime_unset) == -1) { |
|
438 | 448 | goto bail; |
|
439 | 449 | } |
|
440 | 450 | Py_DECREF(mtime_unset); |
|
441 | 451 | mtime_unset = NULL; |
|
442 | 452 | } |
|
443 | 453 | *p++ = state; |
|
444 | 454 | putbe32((uint32_t)mode, p); |
|
445 | 455 | putbe32((uint32_t)size, p + 4); |
|
446 | 456 | putbe32((uint32_t)mtime, p + 8); |
|
447 | 457 | t = p + 12; |
|
448 | 458 | p += 16; |
|
449 | 459 | len = PyBytes_GET_SIZE(k); |
|
450 | 460 | memcpy(p, PyBytes_AS_STRING(k), len); |
|
451 | 461 | p += len; |
|
452 | 462 | o = PyDict_GetItem(copymap, k); |
|
453 | 463 | if (o) { |
|
454 | 464 | *p++ = '\0'; |
|
455 | 465 | l = PyBytes_GET_SIZE(o); |
|
456 | 466 | memcpy(p, PyBytes_AS_STRING(o), l); |
|
457 | 467 | p += l; |
|
458 | 468 | len += l + 1; |
|
459 | 469 | } |
|
460 | 470 | putbe32((uint32_t)len, t); |
|
461 | 471 | } |
|
462 | 472 | |
|
463 | 473 | pos = p - PyBytes_AS_STRING(packobj); |
|
464 | 474 | if (pos != nbytes) { |
|
465 | 475 | PyErr_Format(PyExc_SystemError, "bad dirstate size: %ld != %ld", |
|
466 | 476 | (long)pos, (long)nbytes); |
|
467 | 477 | goto bail; |
|
468 | 478 | } |
|
469 | 479 | |
|
470 | 480 | return packobj; |
|
471 | 481 | bail: |
|
472 | 482 | Py_XDECREF(mtime_unset); |
|
473 | 483 | Py_XDECREF(packobj); |
|
474 | 484 | Py_XDECREF(v); |
|
475 | 485 | return NULL; |
|
476 | 486 | } |
|
477 | 487 | |
|
478 | 488 | #define BUMPED_FIX 1 |
|
479 | 489 | #define USING_SHA_256 2 |
|
480 | 490 | #define FM1_HEADER_SIZE (4 + 8 + 2 + 2 + 1 + 1 + 1) |
|
481 | 491 | |
|
482 | 492 | static PyObject *readshas(const char *source, unsigned char num, |
|
483 | 493 | Py_ssize_t hashwidth) |
|
484 | 494 | { |
|
485 | 495 | int i; |
|
486 | 496 | PyObject *list = PyTuple_New(num); |
|
487 | 497 | if (list == NULL) { |
|
488 | 498 | return NULL; |
|
489 | 499 | } |
|
490 | 500 | for (i = 0; i < num; i++) { |
|
491 | 501 | PyObject *hash = PyBytes_FromStringAndSize(source, hashwidth); |
|
492 | 502 | if (hash == NULL) { |
|
493 | 503 | Py_DECREF(list); |
|
494 | 504 | return NULL; |
|
495 | 505 | } |
|
496 | 506 | PyTuple_SET_ITEM(list, i, hash); |
|
497 | 507 | source += hashwidth; |
|
498 | 508 | } |
|
499 | 509 | return list; |
|
500 | 510 | } |
|
501 | 511 | |
|
502 | 512 | static PyObject *fm1readmarker(const char *databegin, const char *dataend, |
|
503 | 513 | uint32_t *msize) |
|
504 | 514 | { |
|
505 | 515 | const char *data = databegin; |
|
506 | 516 | const char *meta; |
|
507 | 517 | |
|
508 | 518 | double mtime; |
|
509 | 519 | int16_t tz; |
|
510 | 520 | uint16_t flags; |
|
511 | 521 | unsigned char nsuccs, nparents, nmetadata; |
|
512 | 522 | Py_ssize_t hashwidth = 20; |
|
513 | 523 | |
|
514 | 524 | PyObject *prec = NULL, *parents = NULL, *succs = NULL; |
|
515 | 525 | PyObject *metadata = NULL, *ret = NULL; |
|
516 | 526 | int i; |
|
517 | 527 | |
|
518 | 528 | if (data + FM1_HEADER_SIZE > dataend) { |
|
519 | 529 | goto overflow; |
|
520 | 530 | } |
|
521 | 531 | |
|
522 | 532 | *msize = getbe32(data); |
|
523 | 533 | data += 4; |
|
524 | 534 | mtime = getbefloat64(data); |
|
525 | 535 | data += 8; |
|
526 | 536 | tz = getbeint16(data); |
|
527 | 537 | data += 2; |
|
528 | 538 | flags = getbeuint16(data); |
|
529 | 539 | data += 2; |
|
530 | 540 | |
|
531 | 541 | if (flags & USING_SHA_256) { |
|
532 | 542 | hashwidth = 32; |
|
533 | 543 | } |
|
534 | 544 | |
|
535 | 545 | nsuccs = (unsigned char)(*data++); |
|
536 | 546 | nparents = (unsigned char)(*data++); |
|
537 | 547 | nmetadata = (unsigned char)(*data++); |
|
538 | 548 | |
|
539 | 549 | if (databegin + *msize > dataend) { |
|
540 | 550 | goto overflow; |
|
541 | 551 | } |
|
542 | 552 | dataend = databegin + *msize; /* narrow down to marker size */ |
|
543 | 553 | |
|
544 | 554 | if (data + hashwidth > dataend) { |
|
545 | 555 | goto overflow; |
|
546 | 556 | } |
|
547 | 557 | prec = PyBytes_FromStringAndSize(data, hashwidth); |
|
548 | 558 | data += hashwidth; |
|
549 | 559 | if (prec == NULL) { |
|
550 | 560 | goto bail; |
|
551 | 561 | } |
|
552 | 562 | |
|
553 | 563 | if (data + nsuccs * hashwidth > dataend) { |
|
554 | 564 | goto overflow; |
|
555 | 565 | } |
|
556 | 566 | succs = readshas(data, nsuccs, hashwidth); |
|
557 | 567 | if (succs == NULL) { |
|
558 | 568 | goto bail; |
|
559 | 569 | } |
|
560 | 570 | data += nsuccs * hashwidth; |
|
561 | 571 | |
|
562 | 572 | if (nparents == 1 || nparents == 2) { |
|
563 | 573 | if (data + nparents * hashwidth > dataend) { |
|
564 | 574 | goto overflow; |
|
565 | 575 | } |
|
566 | 576 | parents = readshas(data, nparents, hashwidth); |
|
567 | 577 | if (parents == NULL) { |
|
568 | 578 | goto bail; |
|
569 | 579 | } |
|
570 | 580 | data += nparents * hashwidth; |
|
571 | 581 | } else { |
|
572 | 582 | parents = Py_None; |
|
573 | 583 | Py_INCREF(parents); |
|
574 | 584 | } |
|
575 | 585 | |
|
576 | 586 | if (data + 2 * nmetadata > dataend) { |
|
577 | 587 | goto overflow; |
|
578 | 588 | } |
|
579 | 589 | meta = data + (2 * nmetadata); |
|
580 | 590 | metadata = PyTuple_New(nmetadata); |
|
581 | 591 | if (metadata == NULL) { |
|
582 | 592 | goto bail; |
|
583 | 593 | } |
|
584 | 594 | for (i = 0; i < nmetadata; i++) { |
|
585 | 595 | PyObject *tmp, *left = NULL, *right = NULL; |
|
586 | 596 | Py_ssize_t leftsize = (unsigned char)(*data++); |
|
587 | 597 | Py_ssize_t rightsize = (unsigned char)(*data++); |
|
588 | 598 | if (meta + leftsize + rightsize > dataend) { |
|
589 | 599 | goto overflow; |
|
590 | 600 | } |
|
591 | 601 | left = PyBytes_FromStringAndSize(meta, leftsize); |
|
592 | 602 | meta += leftsize; |
|
593 | 603 | right = PyBytes_FromStringAndSize(meta, rightsize); |
|
594 | 604 | meta += rightsize; |
|
595 | 605 | tmp = PyTuple_New(2); |
|
596 | 606 | if (!left || !right || !tmp) { |
|
597 | 607 | Py_XDECREF(left); |
|
598 | 608 | Py_XDECREF(right); |
|
599 | 609 | Py_XDECREF(tmp); |
|
600 | 610 | goto bail; |
|
601 | 611 | } |
|
602 | 612 | PyTuple_SET_ITEM(tmp, 0, left); |
|
603 | 613 | PyTuple_SET_ITEM(tmp, 1, right); |
|
604 | 614 | PyTuple_SET_ITEM(metadata, i, tmp); |
|
605 | 615 | } |
|
606 | 616 | ret = Py_BuildValue("(OOHO(di)O)", prec, succs, flags, metadata, mtime, |
|
607 | 617 | (int)tz * 60, parents); |
|
608 | 618 | goto bail; /* return successfully */ |
|
609 | 619 | |
|
610 | 620 | overflow: |
|
611 | 621 | PyErr_SetString(PyExc_ValueError, "overflow in obsstore"); |
|
612 | 622 | bail: |
|
613 | 623 | Py_XDECREF(prec); |
|
614 | 624 | Py_XDECREF(succs); |
|
615 | 625 | Py_XDECREF(metadata); |
|
616 | 626 | Py_XDECREF(parents); |
|
617 | 627 | return ret; |
|
618 | 628 | } |
|
619 | 629 | |
|
620 | 630 | static PyObject *fm1readmarkers(PyObject *self, PyObject *args) |
|
621 | 631 | { |
|
622 | 632 | const char *data, *dataend; |
|
623 | 633 | Py_ssize_t datalen, offset, stop; |
|
624 | 634 | PyObject *markers = NULL; |
|
625 | 635 | |
|
626 | 636 | if (!PyArg_ParseTuple(args, PY23("s#nn", "y#nn"), &data, &datalen, |
|
627 | 637 | &offset, &stop)) { |
|
628 | 638 | return NULL; |
|
629 | 639 | } |
|
630 | 640 | if (offset < 0) { |
|
631 | 641 | PyErr_SetString(PyExc_ValueError, |
|
632 | 642 | "invalid negative offset in fm1readmarkers"); |
|
633 | 643 | return NULL; |
|
634 | 644 | } |
|
635 | 645 | if (stop > datalen) { |
|
636 | 646 | PyErr_SetString( |
|
637 | 647 | PyExc_ValueError, |
|
638 | 648 | "stop longer than data length in fm1readmarkers"); |
|
639 | 649 | return NULL; |
|
640 | 650 | } |
|
641 | 651 | dataend = data + datalen; |
|
642 | 652 | data += offset; |
|
643 | 653 | markers = PyList_New(0); |
|
644 | 654 | if (!markers) { |
|
645 | 655 | return NULL; |
|
646 | 656 | } |
|
647 | 657 | while (offset < stop) { |
|
648 | 658 | uint32_t msize; |
|
649 | 659 | int error; |
|
650 | 660 | PyObject *record = fm1readmarker(data, dataend, &msize); |
|
651 | 661 | if (!record) { |
|
652 | 662 | goto bail; |
|
653 | 663 | } |
|
654 | 664 | error = PyList_Append(markers, record); |
|
655 | 665 | Py_DECREF(record); |
|
656 | 666 | if (error) { |
|
657 | 667 | goto bail; |
|
658 | 668 | } |
|
659 | 669 | data += msize; |
|
660 | 670 | offset += msize; |
|
661 | 671 | } |
|
662 | 672 | return markers; |
|
663 | 673 | bail: |
|
664 | 674 | Py_DECREF(markers); |
|
665 | 675 | return NULL; |
|
666 | 676 | } |
|
667 | 677 | |
|
668 | 678 | static char parsers_doc[] = "Efficient content parsing."; |
|
669 | 679 | |
|
670 | 680 | PyObject *encodedir(PyObject *self, PyObject *args); |
|
671 | 681 | PyObject *pathencode(PyObject *self, PyObject *args); |
|
672 | 682 | PyObject *lowerencode(PyObject *self, PyObject *args); |
|
673 | 683 | PyObject *parse_index2(PyObject *self, PyObject *args, PyObject *kwargs); |
|
674 | 684 | |
|
675 | 685 | static PyMethodDef methods[] = { |
|
676 | 686 | {"pack_dirstate", pack_dirstate, METH_VARARGS, "pack a dirstate\n"}, |
|
677 | 687 | {"nonnormalotherparententries", nonnormalotherparententries, METH_VARARGS, |
|
678 | 688 | "create a set containing non-normal and other parent entries of given " |
|
679 | 689 | "dirstate\n"}, |
|
680 | 690 | {"parse_dirstate", parse_dirstate, METH_VARARGS, "parse a dirstate\n"}, |
|
681 | 691 | {"parse_index2", (PyCFunction)parse_index2, METH_VARARGS | METH_KEYWORDS, |
|
682 | 692 | "parse a revlog index\n"}, |
|
683 | 693 | {"isasciistr", isasciistr, METH_VARARGS, "check if an ASCII string\n"}, |
|
684 | 694 | {"asciilower", asciilower, METH_VARARGS, "lowercase an ASCII string\n"}, |
|
685 | 695 | {"asciiupper", asciiupper, METH_VARARGS, "uppercase an ASCII string\n"}, |
|
686 | 696 | {"dict_new_presized", dict_new_presized, METH_VARARGS, |
|
687 | 697 | "construct a dict with an expected size\n"}, |
|
688 | 698 | {"make_file_foldmap", make_file_foldmap, METH_VARARGS, |
|
689 | 699 | "make file foldmap\n"}, |
|
690 | 700 | {"jsonescapeu8fast", jsonescapeu8fast, METH_VARARGS, |
|
691 | 701 | "escape a UTF-8 byte string to JSON (fast path)\n"}, |
|
692 | 702 | {"encodedir", encodedir, METH_VARARGS, "encodedir a path\n"}, |
|
693 | 703 | {"pathencode", pathencode, METH_VARARGS, "fncache-encode a path\n"}, |
|
694 | 704 | {"lowerencode", lowerencode, METH_VARARGS, "lower-encode a path\n"}, |
|
695 | 705 | {"fm1readmarkers", fm1readmarkers, METH_VARARGS, |
|
696 | 706 | "parse v1 obsolete markers\n"}, |
|
697 | 707 | {NULL, NULL}}; |
|
698 | 708 | |
|
699 | 709 | void dirs_module_init(PyObject *mod); |
|
700 | 710 | void manifest_module_init(PyObject *mod); |
|
701 | 711 | void revlog_module_init(PyObject *mod); |
|
702 | 712 | |
|
703 | 713 | static const int version = 20; |
|
704 | 714 | |
|
705 | 715 | static void module_init(PyObject *mod) |
|
706 | 716 | { |
|
707 | 717 | PyObject *capsule = NULL; |
|
708 | 718 | PyModule_AddIntConstant(mod, "version", version); |
|
709 | 719 | |
|
710 | 720 | /* This module constant has two purposes. First, it lets us unit test |
|
711 | 721 | * the ImportError raised without hard-coding any error text. This |
|
712 | 722 | * means we can change the text in the future without breaking tests, |
|
713 | 723 | * even across changesets without a recompile. Second, its presence |
|
714 | 724 | * can be used to determine whether the version-checking logic is |
|
715 | 725 | * present, which also helps in testing across changesets without a |
|
716 | 726 | * recompile. Note that this means the pure-Python version of parsers |
|
717 | 727 | * should not have this module constant. */ |
|
718 | 728 | PyModule_AddStringConstant(mod, "versionerrortext", versionerrortext); |
|
719 | 729 | |
|
720 | 730 | dirs_module_init(mod); |
|
721 | 731 | manifest_module_init(mod); |
|
722 | 732 | revlog_module_init(mod); |
|
723 | 733 | |
|
724 | 734 | capsule = PyCapsule_New( |
|
725 | 735 | make_dirstate_tuple, |
|
726 | 736 | "mercurial.cext.parsers.make_dirstate_tuple_CAPI", NULL); |
|
727 | 737 | if (capsule != NULL) |
|
728 | 738 | PyModule_AddObject(mod, "make_dirstate_tuple_CAPI", capsule); |
|
729 | 739 | |
|
730 | 740 | if (PyType_Ready(&dirstateTupleType) < 0) { |
|
731 | 741 | return; |
|
732 | 742 | } |
|
733 | 743 | Py_INCREF(&dirstateTupleType); |
|
734 | 744 | PyModule_AddObject(mod, "dirstatetuple", |
|
735 | 745 | (PyObject *)&dirstateTupleType); |
|
736 | 746 | } |
|
737 | 747 | |
|
738 | 748 | static int check_python_version(void) |
|
739 | 749 | { |
|
740 | 750 | PyObject *sys = PyImport_ImportModule("sys"), *ver; |
|
741 | 751 | long hexversion; |
|
742 | 752 | if (!sys) { |
|
743 | 753 | return -1; |
|
744 | 754 | } |
|
745 | 755 | ver = PyObject_GetAttrString(sys, "hexversion"); |
|
746 | 756 | Py_DECREF(sys); |
|
747 | 757 | if (!ver) { |
|
748 | 758 | return -1; |
|
749 | 759 | } |
|
750 | 760 | hexversion = PyInt_AsLong(ver); |
|
751 | 761 | Py_DECREF(ver); |
|
752 | 762 | /* sys.hexversion is a 32-bit number by default, so the -1 case |
|
753 | 763 | * should only occur in unusual circumstances (e.g. if sys.hexversion |
|
754 | 764 | * is manually set to an invalid value). */ |
|
755 | 765 | if ((hexversion == -1) || (hexversion >> 16 != PY_VERSION_HEX >> 16)) { |
|
756 | 766 | PyErr_Format(PyExc_ImportError, |
|
757 | 767 | "%s: The Mercurial extension " |
|
758 | 768 | "modules were compiled with Python " PY_VERSION |
|
759 | 769 | ", but " |
|
760 | 770 | "Mercurial is currently using Python with " |
|
761 | 771 | "sys.hexversion=%ld: " |
|
762 | 772 | "Python %s\n at: %s", |
|
763 | 773 | versionerrortext, hexversion, Py_GetVersion(), |
|
764 | 774 | Py_GetProgramFullPath()); |
|
765 | 775 | return -1; |
|
766 | 776 | } |
|
767 | 777 | return 0; |
|
768 | 778 | } |
|
769 | 779 | |
|
770 | 780 | #ifdef IS_PY3K |
|
771 | 781 | static struct PyModuleDef parsers_module = {PyModuleDef_HEAD_INIT, "parsers", |
|
772 | 782 | parsers_doc, -1, methods}; |
|
773 | 783 | |
|
774 | 784 | PyMODINIT_FUNC PyInit_parsers(void) |
|
775 | 785 | { |
|
776 | 786 | PyObject *mod; |
|
777 | 787 | |
|
778 | 788 | if (check_python_version() == -1) |
|
779 | 789 | return NULL; |
|
780 | 790 | mod = PyModule_Create(&parsers_module); |
|
781 | 791 | module_init(mod); |
|
782 | 792 | return mod; |
|
783 | 793 | } |
|
784 | 794 | #else |
|
785 | 795 | PyMODINIT_FUNC initparsers(void) |
|
786 | 796 | { |
|
787 | 797 | PyObject *mod; |
|
788 | 798 | |
|
789 | 799 | if (check_python_version() == -1) { |
|
790 | 800 | return; |
|
791 | 801 | } |
|
792 | 802 | mod = Py_InitModule3("parsers", methods, parsers_doc); |
|
793 | 803 | module_init(mod); |
|
794 | 804 | } |
|
795 | 805 | #endif |
@@ -1,1444 +1,1455 b'' | |||
|
1 | 1 | # dirstate.py - working directory tracking for mercurial |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2005-2007 Olivia Mackall <olivia@selenic.com> |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import collections |
|
11 | 11 | import contextlib |
|
12 | 12 | import errno |
|
13 | 13 | import os |
|
14 | 14 | import stat |
|
15 | 15 | |
|
16 | 16 | from .i18n import _ |
|
17 | 17 | from .pycompat import delattr |
|
18 | 18 | |
|
19 | 19 | from hgdemandimport import tracing |
|
20 | 20 | |
|
21 | 21 | from . import ( |
|
22 | 22 | dirstatemap, |
|
23 | 23 | encoding, |
|
24 | 24 | error, |
|
25 | 25 | match as matchmod, |
|
26 | 26 | pathutil, |
|
27 | 27 | policy, |
|
28 | 28 | pycompat, |
|
29 | 29 | scmutil, |
|
30 | 30 | sparse, |
|
31 | 31 | util, |
|
32 | 32 | ) |
|
33 | 33 | |
|
34 | 34 | from .interfaces import ( |
|
35 | 35 | dirstate as intdirstate, |
|
36 | 36 | util as interfaceutil, |
|
37 | 37 | ) |
|
38 | 38 | |
|
39 | 39 | parsers = policy.importmod('parsers') |
|
40 | 40 | rustmod = policy.importrust('dirstate') |
|
41 | 41 | |
|
42 | 42 | SUPPORTS_DIRSTATE_V2 = rustmod is not None |
|
43 | 43 | |
|
44 | 44 | propertycache = util.propertycache |
|
45 | 45 | filecache = scmutil.filecache |
|
46 | 46 | _rangemask = 0x7FFFFFFF |
|
47 | 47 | |
|
48 | 48 | dirstatetuple = parsers.dirstatetuple |
|
49 | 49 | |
|
50 | 50 | |
|
51 | 51 | # a special value used internally for `size` if the file come from the other parent |
|
52 | 52 | FROM_P2 = dirstatemap.FROM_P2 |
|
53 | 53 | |
|
54 | 54 | # a special value used internally for `size` if the file is modified/merged/added |
|
55 | 55 | NONNORMAL = dirstatemap.NONNORMAL |
|
56 | 56 | |
|
57 | 57 | # a special value used internally for `time` if the time is ambigeous |
|
58 | 58 | AMBIGUOUS_TIME = dirstatemap.AMBIGUOUS_TIME |
|
59 | 59 | |
|
60 | 60 | |
|
61 | 61 | class repocache(filecache): |
|
62 | 62 | """filecache for files in .hg/""" |
|
63 | 63 | |
|
64 | 64 | def join(self, obj, fname): |
|
65 | 65 | return obj._opener.join(fname) |
|
66 | 66 | |
|
67 | 67 | |
|
68 | 68 | class rootcache(filecache): |
|
69 | 69 | """filecache for files in the repository root""" |
|
70 | 70 | |
|
71 | 71 | def join(self, obj, fname): |
|
72 | 72 | return obj._join(fname) |
|
73 | 73 | |
|
74 | 74 | |
|
75 | 75 | def _getfsnow(vfs): |
|
76 | 76 | '''Get "now" timestamp on filesystem''' |
|
77 | 77 | tmpfd, tmpname = vfs.mkstemp() |
|
78 | 78 | try: |
|
79 | 79 | return os.fstat(tmpfd)[stat.ST_MTIME] |
|
80 | 80 | finally: |
|
81 | 81 | os.close(tmpfd) |
|
82 | 82 | vfs.unlink(tmpname) |
|
83 | 83 | |
|
84 | 84 | |
|
85 | 85 | @interfaceutil.implementer(intdirstate.idirstate) |
|
86 | 86 | class dirstate(object): |
|
87 | 87 | def __init__( |
|
88 | 88 | self, |
|
89 | 89 | opener, |
|
90 | 90 | ui, |
|
91 | 91 | root, |
|
92 | 92 | validate, |
|
93 | 93 | sparsematchfn, |
|
94 | 94 | nodeconstants, |
|
95 | 95 | use_dirstate_v2, |
|
96 | 96 | ): |
|
97 | 97 | """Create a new dirstate object. |
|
98 | 98 | |
|
99 | 99 | opener is an open()-like callable that can be used to open the |
|
100 | 100 | dirstate file; root is the root of the directory tracked by |
|
101 | 101 | the dirstate. |
|
102 | 102 | """ |
|
103 | 103 | self._use_dirstate_v2 = use_dirstate_v2 |
|
104 | 104 | self._nodeconstants = nodeconstants |
|
105 | 105 | self._opener = opener |
|
106 | 106 | self._validate = validate |
|
107 | 107 | self._root = root |
|
108 | 108 | self._sparsematchfn = sparsematchfn |
|
109 | 109 | # ntpath.join(root, '') of Python 2.7.9 does not add sep if root is |
|
110 | 110 | # UNC path pointing to root share (issue4557) |
|
111 | 111 | self._rootdir = pathutil.normasprefix(root) |
|
112 | 112 | self._dirty = False |
|
113 | 113 | self._lastnormaltime = 0 |
|
114 | 114 | self._ui = ui |
|
115 | 115 | self._filecache = {} |
|
116 | 116 | self._parentwriters = 0 |
|
117 | 117 | self._filename = b'dirstate' |
|
118 | 118 | self._pendingfilename = b'%s.pending' % self._filename |
|
119 | 119 | self._plchangecallbacks = {} |
|
120 | 120 | self._origpl = None |
|
121 | 121 | self._updatedfiles = set() |
|
122 | 122 | self._mapcls = dirstatemap.dirstatemap |
|
123 | 123 | # Access and cache cwd early, so we don't access it for the first time |
|
124 | 124 | # after a working-copy update caused it to not exist (accessing it then |
|
125 | 125 | # raises an exception). |
|
126 | 126 | self._cwd |
|
127 | 127 | |
|
128 | 128 | def prefetch_parents(self): |
|
129 | 129 | """make sure the parents are loaded |
|
130 | 130 | |
|
131 | 131 | Used to avoid a race condition. |
|
132 | 132 | """ |
|
133 | 133 | self._pl |
|
134 | 134 | |
|
135 | 135 | @contextlib.contextmanager |
|
136 | 136 | def parentchange(self): |
|
137 | 137 | """Context manager for handling dirstate parents. |
|
138 | 138 | |
|
139 | 139 | If an exception occurs in the scope of the context manager, |
|
140 | 140 | the incoherent dirstate won't be written when wlock is |
|
141 | 141 | released. |
|
142 | 142 | """ |
|
143 | 143 | self._parentwriters += 1 |
|
144 | 144 | yield |
|
145 | 145 | # Typically we want the "undo" step of a context manager in a |
|
146 | 146 | # finally block so it happens even when an exception |
|
147 | 147 | # occurs. In this case, however, we only want to decrement |
|
148 | 148 | # parentwriters if the code in the with statement exits |
|
149 | 149 | # normally, so we don't have a try/finally here on purpose. |
|
150 | 150 | self._parentwriters -= 1 |
|
151 | 151 | |
|
152 | 152 | def pendingparentchange(self): |
|
153 | 153 | """Returns true if the dirstate is in the middle of a set of changes |
|
154 | 154 | that modify the dirstate parent. |
|
155 | 155 | """ |
|
156 | 156 | return self._parentwriters > 0 |
|
157 | 157 | |
|
158 | 158 | @propertycache |
|
159 | 159 | def _map(self): |
|
160 | 160 | """Return the dirstate contents (see documentation for dirstatemap).""" |
|
161 | 161 | self._map = self._mapcls( |
|
162 | 162 | self._ui, |
|
163 | 163 | self._opener, |
|
164 | 164 | self._root, |
|
165 | 165 | self._nodeconstants, |
|
166 | 166 | self._use_dirstate_v2, |
|
167 | 167 | ) |
|
168 | 168 | return self._map |
|
169 | 169 | |
|
170 | 170 | @property |
|
171 | 171 | def _sparsematcher(self): |
|
172 | 172 | """The matcher for the sparse checkout. |
|
173 | 173 | |
|
174 | 174 | The working directory may not include every file from a manifest. The |
|
175 | 175 | matcher obtained by this property will match a path if it is to be |
|
176 | 176 | included in the working directory. |
|
177 | 177 | """ |
|
178 | 178 | # TODO there is potential to cache this property. For now, the matcher |
|
179 | 179 | # is resolved on every access. (But the called function does use a |
|
180 | 180 | # cache to keep the lookup fast.) |
|
181 | 181 | return self._sparsematchfn() |
|
182 | 182 | |
|
183 | 183 | @repocache(b'branch') |
|
184 | 184 | def _branch(self): |
|
185 | 185 | try: |
|
186 | 186 | return self._opener.read(b"branch").strip() or b"default" |
|
187 | 187 | except IOError as inst: |
|
188 | 188 | if inst.errno != errno.ENOENT: |
|
189 | 189 | raise |
|
190 | 190 | return b"default" |
|
191 | 191 | |
|
192 | 192 | @property |
|
193 | 193 | def _pl(self): |
|
194 | 194 | return self._map.parents() |
|
195 | 195 | |
|
196 | 196 | def hasdir(self, d): |
|
197 | 197 | return self._map.hastrackeddir(d) |
|
198 | 198 | |
|
199 | 199 | @rootcache(b'.hgignore') |
|
200 | 200 | def _ignore(self): |
|
201 | 201 | files = self._ignorefiles() |
|
202 | 202 | if not files: |
|
203 | 203 | return matchmod.never() |
|
204 | 204 | |
|
205 | 205 | pats = [b'include:%s' % f for f in files] |
|
206 | 206 | return matchmod.match(self._root, b'', [], pats, warn=self._ui.warn) |
|
207 | 207 | |
|
208 | 208 | @propertycache |
|
209 | 209 | def _slash(self): |
|
210 | 210 | return self._ui.configbool(b'ui', b'slash') and pycompat.ossep != b'/' |
|
211 | 211 | |
|
212 | 212 | @propertycache |
|
213 | 213 | def _checklink(self): |
|
214 | 214 | return util.checklink(self._root) |
|
215 | 215 | |
|
216 | 216 | @propertycache |
|
217 | 217 | def _checkexec(self): |
|
218 | 218 | return bool(util.checkexec(self._root)) |
|
219 | 219 | |
|
220 | 220 | @propertycache |
|
221 | 221 | def _checkcase(self): |
|
222 | 222 | return not util.fscasesensitive(self._join(b'.hg')) |
|
223 | 223 | |
|
224 | 224 | def _join(self, f): |
|
225 | 225 | # much faster than os.path.join() |
|
226 | 226 | # it's safe because f is always a relative path |
|
227 | 227 | return self._rootdir + f |
|
228 | 228 | |
|
229 | 229 | def flagfunc(self, buildfallback): |
|
230 | 230 | if self._checklink and self._checkexec: |
|
231 | 231 | |
|
232 | 232 | def f(x): |
|
233 | 233 | try: |
|
234 | 234 | st = os.lstat(self._join(x)) |
|
235 | 235 | if util.statislink(st): |
|
236 | 236 | return b'l' |
|
237 | 237 | if util.statisexec(st): |
|
238 | 238 | return b'x' |
|
239 | 239 | except OSError: |
|
240 | 240 | pass |
|
241 | 241 | return b'' |
|
242 | 242 | |
|
243 | 243 | return f |
|
244 | 244 | |
|
245 | 245 | fallback = buildfallback() |
|
246 | 246 | if self._checklink: |
|
247 | 247 | |
|
248 | 248 | def f(x): |
|
249 | 249 | if os.path.islink(self._join(x)): |
|
250 | 250 | return b'l' |
|
251 | 251 | if b'x' in fallback(x): |
|
252 | 252 | return b'x' |
|
253 | 253 | return b'' |
|
254 | 254 | |
|
255 | 255 | return f |
|
256 | 256 | if self._checkexec: |
|
257 | 257 | |
|
258 | 258 | def f(x): |
|
259 | 259 | if b'l' in fallback(x): |
|
260 | 260 | return b'l' |
|
261 | 261 | if util.isexec(self._join(x)): |
|
262 | 262 | return b'x' |
|
263 | 263 | return b'' |
|
264 | 264 | |
|
265 | 265 | return f |
|
266 | 266 | else: |
|
267 | 267 | return fallback |
|
268 | 268 | |
|
269 | 269 | @propertycache |
|
270 | 270 | def _cwd(self): |
|
271 | 271 | # internal config: ui.forcecwd |
|
272 | 272 | forcecwd = self._ui.config(b'ui', b'forcecwd') |
|
273 | 273 | if forcecwd: |
|
274 | 274 | return forcecwd |
|
275 | 275 | return encoding.getcwd() |
|
276 | 276 | |
|
277 | 277 | def getcwd(self): |
|
278 | 278 | """Return the path from which a canonical path is calculated. |
|
279 | 279 | |
|
280 | 280 | This path should be used to resolve file patterns or to convert |
|
281 | 281 | canonical paths back to file paths for display. It shouldn't be |
|
282 | 282 | used to get real file paths. Use vfs functions instead. |
|
283 | 283 | """ |
|
284 | 284 | cwd = self._cwd |
|
285 | 285 | if cwd == self._root: |
|
286 | 286 | return b'' |
|
287 | 287 | # self._root ends with a path separator if self._root is '/' or 'C:\' |
|
288 | 288 | rootsep = self._root |
|
289 | 289 | if not util.endswithsep(rootsep): |
|
290 | 290 | rootsep += pycompat.ossep |
|
291 | 291 | if cwd.startswith(rootsep): |
|
292 | 292 | return cwd[len(rootsep) :] |
|
293 | 293 | else: |
|
294 | 294 | # we're outside the repo. return an absolute path. |
|
295 | 295 | return cwd |
|
296 | 296 | |
|
297 | 297 | def pathto(self, f, cwd=None): |
|
298 | 298 | if cwd is None: |
|
299 | 299 | cwd = self.getcwd() |
|
300 | 300 | path = util.pathto(self._root, cwd, f) |
|
301 | 301 | if self._slash: |
|
302 | 302 | return util.pconvert(path) |
|
303 | 303 | return path |
|
304 | 304 | |
|
305 | 305 | def __getitem__(self, key): |
|
306 | 306 | """Return the current state of key (a filename) in the dirstate. |
|
307 | 307 | |
|
308 | 308 | States are: |
|
309 | 309 | n normal |
|
310 | 310 | m needs merging |
|
311 | 311 | r marked for removal |
|
312 | 312 | a marked for addition |
|
313 | 313 | ? not tracked |
|
314 | ||
|
315 | XXX The "state" is a bit obscure to be in the "public" API. we should | |
|
316 | consider migrating all user of this to going through the dirstate entry | |
|
317 | instead. | |
|
314 | 318 | """ |
|
315 |
|
|
|
319 | entry = self._map.get(key) | |
|
320 | if entry is not None: | |
|
321 | return entry.state | |
|
322 | return b'?' | |
|
316 | 323 | |
|
317 | 324 | def __contains__(self, key): |
|
318 | 325 | return key in self._map |
|
319 | 326 | |
|
320 | 327 | def __iter__(self): |
|
321 | 328 | return iter(sorted(self._map)) |
|
322 | 329 | |
|
323 | 330 | def items(self): |
|
324 | 331 | return pycompat.iteritems(self._map) |
|
325 | 332 | |
|
326 | 333 | iteritems = items |
|
327 | 334 | |
|
328 | 335 | def directories(self): |
|
329 | 336 | return self._map.directories() |
|
330 | 337 | |
|
331 | 338 | def parents(self): |
|
332 | 339 | return [self._validate(p) for p in self._pl] |
|
333 | 340 | |
|
334 | 341 | def p1(self): |
|
335 | 342 | return self._validate(self._pl[0]) |
|
336 | 343 | |
|
337 | 344 | def p2(self): |
|
338 | 345 | return self._validate(self._pl[1]) |
|
339 | 346 | |
|
340 | 347 | @property |
|
341 | 348 | def in_merge(self): |
|
342 | 349 | """True if a merge is in progress""" |
|
343 | 350 | return self._pl[1] != self._nodeconstants.nullid |
|
344 | 351 | |
|
345 | 352 | def branch(self): |
|
346 | 353 | return encoding.tolocal(self._branch) |
|
347 | 354 | |
|
348 | 355 | def setparents(self, p1, p2=None): |
|
349 | 356 | """Set dirstate parents to p1 and p2. |
|
350 | 357 | |
|
351 | 358 | When moving from two parents to one, 'm' merged entries a |
|
352 | 359 | adjusted to normal and previous copy records discarded and |
|
353 | 360 | returned by the call. |
|
354 | 361 | |
|
355 | 362 | See localrepo.setparents() |
|
356 | 363 | """ |
|
357 | 364 | if p2 is None: |
|
358 | 365 | p2 = self._nodeconstants.nullid |
|
359 | 366 | if self._parentwriters == 0: |
|
360 | 367 | raise ValueError( |
|
361 | 368 | b"cannot set dirstate parent outside of " |
|
362 | 369 | b"dirstate.parentchange context manager" |
|
363 | 370 | ) |
|
364 | 371 | |
|
365 | 372 | self._dirty = True |
|
366 | 373 | oldp2 = self._pl[1] |
|
367 | 374 | if self._origpl is None: |
|
368 | 375 | self._origpl = self._pl |
|
369 | 376 | self._map.setparents(p1, p2) |
|
370 | 377 | copies = {} |
|
371 | 378 | if ( |
|
372 | 379 | oldp2 != self._nodeconstants.nullid |
|
373 | 380 | and p2 == self._nodeconstants.nullid |
|
374 | 381 | ): |
|
375 | 382 | candidatefiles = self._map.non_normal_or_other_parent_paths() |
|
376 | 383 | |
|
377 | 384 | for f in candidatefiles: |
|
378 | 385 | s = self._map.get(f) |
|
379 | 386 | if s is None: |
|
380 | 387 | continue |
|
381 | 388 | |
|
382 | 389 | # Discard 'm' markers when moving away from a merge state |
|
383 |
if s |
|
|
390 | if s.state == b'm': | |
|
384 | 391 | source = self._map.copymap.get(f) |
|
385 | 392 | if source: |
|
386 | 393 | copies[f] = source |
|
387 | 394 | self.normallookup(f) |
|
388 | 395 | # Also fix up otherparent markers |
|
389 |
elif s |
|
|
396 | elif s.state == b'n' and s[2] == FROM_P2: | |
|
390 | 397 | source = self._map.copymap.get(f) |
|
391 | 398 | if source: |
|
392 | 399 | copies[f] = source |
|
393 | 400 | self.add(f) |
|
394 | 401 | return copies |
|
395 | 402 | |
|
396 | 403 | def setbranch(self, branch): |
|
397 | 404 | self.__class__._branch.set(self, encoding.fromlocal(branch)) |
|
398 | 405 | f = self._opener(b'branch', b'w', atomictemp=True, checkambig=True) |
|
399 | 406 | try: |
|
400 | 407 | f.write(self._branch + b'\n') |
|
401 | 408 | f.close() |
|
402 | 409 | |
|
403 | 410 | # make sure filecache has the correct stat info for _branch after |
|
404 | 411 | # replacing the underlying file |
|
405 | 412 | ce = self._filecache[b'_branch'] |
|
406 | 413 | if ce: |
|
407 | 414 | ce.refresh() |
|
408 | 415 | except: # re-raises |
|
409 | 416 | f.discard() |
|
410 | 417 | raise |
|
411 | 418 | |
|
412 | 419 | def invalidate(self): |
|
413 | 420 | """Causes the next access to reread the dirstate. |
|
414 | 421 | |
|
415 | 422 | This is different from localrepo.invalidatedirstate() because it always |
|
416 | 423 | rereads the dirstate. Use localrepo.invalidatedirstate() if you want to |
|
417 | 424 | check whether the dirstate has changed before rereading it.""" |
|
418 | 425 | |
|
419 | 426 | for a in ("_map", "_branch", "_ignore"): |
|
420 | 427 | if a in self.__dict__: |
|
421 | 428 | delattr(self, a) |
|
422 | 429 | self._lastnormaltime = 0 |
|
423 | 430 | self._dirty = False |
|
424 | 431 | self._updatedfiles.clear() |
|
425 | 432 | self._parentwriters = 0 |
|
426 | 433 | self._origpl = None |
|
427 | 434 | |
|
428 | 435 | def copy(self, source, dest): |
|
429 | 436 | """Mark dest as a copy of source. Unmark dest if source is None.""" |
|
430 | 437 | if source == dest: |
|
431 | 438 | return |
|
432 | 439 | self._dirty = True |
|
433 | 440 | if source is not None: |
|
434 | 441 | self._map.copymap[dest] = source |
|
435 | 442 | self._updatedfiles.add(source) |
|
436 | 443 | self._updatedfiles.add(dest) |
|
437 | 444 | elif self._map.copymap.pop(dest, None): |
|
438 | 445 | self._updatedfiles.add(dest) |
|
439 | 446 | |
|
440 | 447 | def copied(self, file): |
|
441 | 448 | return self._map.copymap.get(file, None) |
|
442 | 449 | |
|
443 | 450 | def copies(self): |
|
444 | 451 | return self._map.copymap |
|
445 | 452 | |
|
446 | 453 | def _addpath( |
|
447 | 454 | self, |
|
448 | 455 | f, |
|
449 | 456 | state, |
|
450 | 457 | mode, |
|
451 | 458 | size=NONNORMAL, |
|
452 | 459 | mtime=AMBIGUOUS_TIME, |
|
453 | 460 | from_p2=False, |
|
454 | 461 | possibly_dirty=False, |
|
455 | 462 | ): |
|
456 | 463 | oldstate = self[f] |
|
457 | 464 | if state == b'a' or oldstate == b'r': |
|
458 | 465 | scmutil.checkfilename(f) |
|
459 | 466 | if self._map.hastrackeddir(f): |
|
460 | 467 | msg = _(b'directory %r already in dirstate') |
|
461 | 468 | msg %= pycompat.bytestr(f) |
|
462 | 469 | raise error.Abort(msg) |
|
463 | 470 | # shadows |
|
464 | 471 | for d in pathutil.finddirs(f): |
|
465 | 472 | if self._map.hastrackeddir(d): |
|
466 | 473 | break |
|
467 | 474 | entry = self._map.get(d) |
|
468 |
if entry is not None and entry |
|
|
475 | if entry is not None and entry.state != b'r': | |
|
469 | 476 | msg = _(b'file %r in dirstate clashes with %r') |
|
470 | 477 | msg %= (pycompat.bytestr(d), pycompat.bytestr(f)) |
|
471 | 478 | raise error.Abort(msg) |
|
472 | 479 | if state == b'a': |
|
473 | 480 | assert not possibly_dirty |
|
474 | 481 | assert not from_p2 |
|
475 | 482 | size = NONNORMAL |
|
476 | 483 | mtime = AMBIGUOUS_TIME |
|
477 | 484 | elif from_p2: |
|
478 | 485 | assert not possibly_dirty |
|
479 | 486 | size = FROM_P2 |
|
480 | 487 | mtime = AMBIGUOUS_TIME |
|
481 | 488 | elif possibly_dirty: |
|
482 | 489 | mtime = AMBIGUOUS_TIME |
|
483 | 490 | else: |
|
484 | 491 | assert size != FROM_P2 |
|
485 | 492 | assert size != NONNORMAL |
|
486 | 493 | size = size & _rangemask |
|
487 | 494 | mtime = mtime & _rangemask |
|
488 | 495 | self._dirty = True |
|
489 | 496 | self._updatedfiles.add(f) |
|
490 | 497 | self._map.addfile(f, oldstate, state, mode, size, mtime) |
|
491 | 498 | |
|
492 | 499 | def normal(self, f, parentfiledata=None): |
|
493 | 500 | """Mark a file normal and clean. |
|
494 | 501 | |
|
495 | 502 | parentfiledata: (mode, size, mtime) of the clean file |
|
496 | 503 | |
|
497 | 504 | parentfiledata should be computed from memory (for mode, |
|
498 | 505 | size), as or close as possible from the point where we |
|
499 | 506 | determined the file was clean, to limit the risk of the |
|
500 | 507 | file having been changed by an external process between the |
|
501 | 508 | moment where the file was determined to be clean and now.""" |
|
502 | 509 | if parentfiledata: |
|
503 | 510 | (mode, size, mtime) = parentfiledata |
|
504 | 511 | else: |
|
505 | 512 | s = os.lstat(self._join(f)) |
|
506 | 513 | mode = s.st_mode |
|
507 | 514 | size = s.st_size |
|
508 | 515 | mtime = s[stat.ST_MTIME] |
|
509 | 516 | self._addpath(f, b'n', mode, size, mtime) |
|
510 | 517 | self._map.copymap.pop(f, None) |
|
511 | 518 | if f in self._map.nonnormalset: |
|
512 | 519 | self._map.nonnormalset.remove(f) |
|
513 | 520 | if mtime > self._lastnormaltime: |
|
514 | 521 | # Remember the most recent modification timeslot for status(), |
|
515 | 522 | # to make sure we won't miss future size-preserving file content |
|
516 | 523 | # modifications that happen within the same timeslot. |
|
517 | 524 | self._lastnormaltime = mtime |
|
518 | 525 | |
|
519 | 526 | def normallookup(self, f): |
|
520 | 527 | '''Mark a file normal, but possibly dirty.''' |
|
521 | 528 | if self.in_merge: |
|
522 | 529 | # if there is a merge going on and the file was either |
|
523 | 530 | # in state 'm' (-1) or coming from other parent (-2) before |
|
524 | 531 | # being removed, restore that state. |
|
525 | 532 | entry = self._map.get(f) |
|
526 | 533 | if entry is not None: |
|
527 |
if entry |
|
|
534 | if entry.state == b'r' and entry[2] in (NONNORMAL, FROM_P2): | |
|
528 | 535 | source = self._map.copymap.get(f) |
|
529 | 536 | if entry[2] == NONNORMAL: |
|
530 | 537 | self.merge(f) |
|
531 | 538 | elif entry[2] == FROM_P2: |
|
532 | 539 | self.otherparent(f) |
|
533 | 540 | if source: |
|
534 | 541 | self.copy(source, f) |
|
535 | 542 | return |
|
536 | if entry[0] == b'm' or entry[0] == b'n' and entry[2] == FROM_P2: | |
|
543 | if ( | |
|
544 | entry.state == b'm' | |
|
545 | or entry.state == b'n' | |
|
546 | and entry[2] == FROM_P2 | |
|
547 | ): | |
|
537 | 548 | return |
|
538 | 549 | self._addpath(f, b'n', 0, possibly_dirty=True) |
|
539 | 550 | self._map.copymap.pop(f, None) |
|
540 | 551 | |
|
541 | 552 | def otherparent(self, f): |
|
542 | 553 | '''Mark as coming from the other parent, always dirty.''' |
|
543 | 554 | if not self.in_merge: |
|
544 | 555 | msg = _(b"setting %r to other parent only allowed in merges") % f |
|
545 | 556 | raise error.Abort(msg) |
|
546 | 557 | if f in self and self[f] == b'n': |
|
547 | 558 | # merge-like |
|
548 | 559 | self._addpath(f, b'm', 0, from_p2=True) |
|
549 | 560 | else: |
|
550 | 561 | # add-like |
|
551 | 562 | self._addpath(f, b'n', 0, from_p2=True) |
|
552 | 563 | self._map.copymap.pop(f, None) |
|
553 | 564 | |
|
554 | 565 | def add(self, f): |
|
555 | 566 | '''Mark a file added.''' |
|
556 | 567 | self._addpath(f, b'a', 0) |
|
557 | 568 | self._map.copymap.pop(f, None) |
|
558 | 569 | |
|
559 | 570 | def remove(self, f): |
|
560 | 571 | '''Mark a file removed.''' |
|
561 | 572 | self._dirty = True |
|
562 | 573 | self._updatedfiles.add(f) |
|
563 | 574 | self._map.removefile(f, in_merge=self.in_merge) |
|
564 | 575 | |
|
565 | 576 | def merge(self, f): |
|
566 | 577 | '''Mark a file merged.''' |
|
567 | 578 | if not self.in_merge: |
|
568 | 579 | return self.normallookup(f) |
|
569 | 580 | return self.otherparent(f) |
|
570 | 581 | |
|
571 | 582 | def drop(self, f): |
|
572 | 583 | '''Drop a file from the dirstate''' |
|
573 | 584 | oldstate = self[f] |
|
574 | 585 | if self._map.dropfile(f, oldstate): |
|
575 | 586 | self._dirty = True |
|
576 | 587 | self._updatedfiles.add(f) |
|
577 | 588 | self._map.copymap.pop(f, None) |
|
578 | 589 | |
|
579 | 590 | def _discoverpath(self, path, normed, ignoremissing, exists, storemap): |
|
580 | 591 | if exists is None: |
|
581 | 592 | exists = os.path.lexists(os.path.join(self._root, path)) |
|
582 | 593 | if not exists: |
|
583 | 594 | # Maybe a path component exists |
|
584 | 595 | if not ignoremissing and b'/' in path: |
|
585 | 596 | d, f = path.rsplit(b'/', 1) |
|
586 | 597 | d = self._normalize(d, False, ignoremissing, None) |
|
587 | 598 | folded = d + b"/" + f |
|
588 | 599 | else: |
|
589 | 600 | # No path components, preserve original case |
|
590 | 601 | folded = path |
|
591 | 602 | else: |
|
592 | 603 | # recursively normalize leading directory components |
|
593 | 604 | # against dirstate |
|
594 | 605 | if b'/' in normed: |
|
595 | 606 | d, f = normed.rsplit(b'/', 1) |
|
596 | 607 | d = self._normalize(d, False, ignoremissing, True) |
|
597 | 608 | r = self._root + b"/" + d |
|
598 | 609 | folded = d + b"/" + util.fspath(f, r) |
|
599 | 610 | else: |
|
600 | 611 | folded = util.fspath(normed, self._root) |
|
601 | 612 | storemap[normed] = folded |
|
602 | 613 | |
|
603 | 614 | return folded |
|
604 | 615 | |
|
605 | 616 | def _normalizefile(self, path, isknown, ignoremissing=False, exists=None): |
|
606 | 617 | normed = util.normcase(path) |
|
607 | 618 | folded = self._map.filefoldmap.get(normed, None) |
|
608 | 619 | if folded is None: |
|
609 | 620 | if isknown: |
|
610 | 621 | folded = path |
|
611 | 622 | else: |
|
612 | 623 | folded = self._discoverpath( |
|
613 | 624 | path, normed, ignoremissing, exists, self._map.filefoldmap |
|
614 | 625 | ) |
|
615 | 626 | return folded |
|
616 | 627 | |
|
617 | 628 | def _normalize(self, path, isknown, ignoremissing=False, exists=None): |
|
618 | 629 | normed = util.normcase(path) |
|
619 | 630 | folded = self._map.filefoldmap.get(normed, None) |
|
620 | 631 | if folded is None: |
|
621 | 632 | folded = self._map.dirfoldmap.get(normed, None) |
|
622 | 633 | if folded is None: |
|
623 | 634 | if isknown: |
|
624 | 635 | folded = path |
|
625 | 636 | else: |
|
626 | 637 | # store discovered result in dirfoldmap so that future |
|
627 | 638 | # normalizefile calls don't start matching directories |
|
628 | 639 | folded = self._discoverpath( |
|
629 | 640 | path, normed, ignoremissing, exists, self._map.dirfoldmap |
|
630 | 641 | ) |
|
631 | 642 | return folded |
|
632 | 643 | |
|
633 | 644 | def normalize(self, path, isknown=False, ignoremissing=False): |
|
634 | 645 | """ |
|
635 | 646 | normalize the case of a pathname when on a casefolding filesystem |
|
636 | 647 | |
|
637 | 648 | isknown specifies whether the filename came from walking the |
|
638 | 649 | disk, to avoid extra filesystem access. |
|
639 | 650 | |
|
640 | 651 | If ignoremissing is True, missing path are returned |
|
641 | 652 | unchanged. Otherwise, we try harder to normalize possibly |
|
642 | 653 | existing path components. |
|
643 | 654 | |
|
644 | 655 | The normalized case is determined based on the following precedence: |
|
645 | 656 | |
|
646 | 657 | - version of name already stored in the dirstate |
|
647 | 658 | - version of name stored on disk |
|
648 | 659 | - version provided via command arguments |
|
649 | 660 | """ |
|
650 | 661 | |
|
651 | 662 | if self._checkcase: |
|
652 | 663 | return self._normalize(path, isknown, ignoremissing) |
|
653 | 664 | return path |
|
654 | 665 | |
|
655 | 666 | def clear(self): |
|
656 | 667 | self._map.clear() |
|
657 | 668 | self._lastnormaltime = 0 |
|
658 | 669 | self._updatedfiles.clear() |
|
659 | 670 | self._dirty = True |
|
660 | 671 | |
|
661 | 672 | def rebuild(self, parent, allfiles, changedfiles=None): |
|
662 | 673 | if changedfiles is None: |
|
663 | 674 | # Rebuild entire dirstate |
|
664 | 675 | to_lookup = allfiles |
|
665 | 676 | to_drop = [] |
|
666 | 677 | lastnormaltime = self._lastnormaltime |
|
667 | 678 | self.clear() |
|
668 | 679 | self._lastnormaltime = lastnormaltime |
|
669 | 680 | elif len(changedfiles) < 10: |
|
670 | 681 | # Avoid turning allfiles into a set, which can be expensive if it's |
|
671 | 682 | # large. |
|
672 | 683 | to_lookup = [] |
|
673 | 684 | to_drop = [] |
|
674 | 685 | for f in changedfiles: |
|
675 | 686 | if f in allfiles: |
|
676 | 687 | to_lookup.append(f) |
|
677 | 688 | else: |
|
678 | 689 | to_drop.append(f) |
|
679 | 690 | else: |
|
680 | 691 | changedfilesset = set(changedfiles) |
|
681 | 692 | to_lookup = changedfilesset & set(allfiles) |
|
682 | 693 | to_drop = changedfilesset - to_lookup |
|
683 | 694 | |
|
684 | 695 | if self._origpl is None: |
|
685 | 696 | self._origpl = self._pl |
|
686 | 697 | self._map.setparents(parent, self._nodeconstants.nullid) |
|
687 | 698 | |
|
688 | 699 | for f in to_lookup: |
|
689 | 700 | self.normallookup(f) |
|
690 | 701 | for f in to_drop: |
|
691 | 702 | self.drop(f) |
|
692 | 703 | |
|
693 | 704 | self._dirty = True |
|
694 | 705 | |
|
695 | 706 | def identity(self): |
|
696 | 707 | """Return identity of dirstate itself to detect changing in storage |
|
697 | 708 | |
|
698 | 709 | If identity of previous dirstate is equal to this, writing |
|
699 | 710 | changes based on the former dirstate out can keep consistency. |
|
700 | 711 | """ |
|
701 | 712 | return self._map.identity |
|
702 | 713 | |
|
703 | 714 | def write(self, tr): |
|
704 | 715 | if not self._dirty: |
|
705 | 716 | return |
|
706 | 717 | |
|
707 | 718 | filename = self._filename |
|
708 | 719 | if tr: |
|
709 | 720 | # 'dirstate.write()' is not only for writing in-memory |
|
710 | 721 | # changes out, but also for dropping ambiguous timestamp. |
|
711 | 722 | # delayed writing re-raise "ambiguous timestamp issue". |
|
712 | 723 | # See also the wiki page below for detail: |
|
713 | 724 | # https://www.mercurial-scm.org/wiki/DirstateTransactionPlan |
|
714 | 725 | |
|
715 | 726 | # emulate dropping timestamp in 'parsers.pack_dirstate' |
|
716 | 727 | now = _getfsnow(self._opener) |
|
717 | 728 | self._map.clearambiguoustimes(self._updatedfiles, now) |
|
718 | 729 | |
|
719 | 730 | # emulate that all 'dirstate.normal' results are written out |
|
720 | 731 | self._lastnormaltime = 0 |
|
721 | 732 | self._updatedfiles.clear() |
|
722 | 733 | |
|
723 | 734 | # delay writing in-memory changes out |
|
724 | 735 | tr.addfilegenerator( |
|
725 | 736 | b'dirstate', |
|
726 | 737 | (self._filename,), |
|
727 | 738 | self._writedirstate, |
|
728 | 739 | location=b'plain', |
|
729 | 740 | ) |
|
730 | 741 | return |
|
731 | 742 | |
|
732 | 743 | st = self._opener(filename, b"w", atomictemp=True, checkambig=True) |
|
733 | 744 | self._writedirstate(st) |
|
734 | 745 | |
|
735 | 746 | def addparentchangecallback(self, category, callback): |
|
736 | 747 | """add a callback to be called when the wd parents are changed |
|
737 | 748 | |
|
738 | 749 | Callback will be called with the following arguments: |
|
739 | 750 | dirstate, (oldp1, oldp2), (newp1, newp2) |
|
740 | 751 | |
|
741 | 752 | Category is a unique identifier to allow overwriting an old callback |
|
742 | 753 | with a newer callback. |
|
743 | 754 | """ |
|
744 | 755 | self._plchangecallbacks[category] = callback |
|
745 | 756 | |
|
746 | 757 | def _writedirstate(self, st): |
|
747 | 758 | # notify callbacks about parents change |
|
748 | 759 | if self._origpl is not None and self._origpl != self._pl: |
|
749 | 760 | for c, callback in sorted( |
|
750 | 761 | pycompat.iteritems(self._plchangecallbacks) |
|
751 | 762 | ): |
|
752 | 763 | callback(self, self._origpl, self._pl) |
|
753 | 764 | self._origpl = None |
|
754 | 765 | # use the modification time of the newly created temporary file as the |
|
755 | 766 | # filesystem's notion of 'now' |
|
756 | 767 | now = util.fstat(st)[stat.ST_MTIME] & _rangemask |
|
757 | 768 | |
|
758 | 769 | # enough 'delaywrite' prevents 'pack_dirstate' from dropping |
|
759 | 770 | # timestamp of each entries in dirstate, because of 'now > mtime' |
|
760 | 771 | delaywrite = self._ui.configint(b'debug', b'dirstate.delaywrite') |
|
761 | 772 | if delaywrite > 0: |
|
762 | 773 | # do we have any files to delay for? |
|
763 | 774 | for f, e in pycompat.iteritems(self._map): |
|
764 |
if e |
|
|
775 | if e.state == b'n' and e[3] == now: | |
|
765 | 776 | import time # to avoid useless import |
|
766 | 777 | |
|
767 | 778 | # rather than sleep n seconds, sleep until the next |
|
768 | 779 | # multiple of n seconds |
|
769 | 780 | clock = time.time() |
|
770 | 781 | start = int(clock) - (int(clock) % delaywrite) |
|
771 | 782 | end = start + delaywrite |
|
772 | 783 | time.sleep(end - clock) |
|
773 | 784 | now = end # trust our estimate that the end is near now |
|
774 | 785 | break |
|
775 | 786 | |
|
776 | 787 | self._map.write(st, now) |
|
777 | 788 | self._lastnormaltime = 0 |
|
778 | 789 | self._dirty = False |
|
779 | 790 | |
|
780 | 791 | def _dirignore(self, f): |
|
781 | 792 | if self._ignore(f): |
|
782 | 793 | return True |
|
783 | 794 | for p in pathutil.finddirs(f): |
|
784 | 795 | if self._ignore(p): |
|
785 | 796 | return True |
|
786 | 797 | return False |
|
787 | 798 | |
|
788 | 799 | def _ignorefiles(self): |
|
789 | 800 | files = [] |
|
790 | 801 | if os.path.exists(self._join(b'.hgignore')): |
|
791 | 802 | files.append(self._join(b'.hgignore')) |
|
792 | 803 | for name, path in self._ui.configitems(b"ui"): |
|
793 | 804 | if name == b'ignore' or name.startswith(b'ignore.'): |
|
794 | 805 | # we need to use os.path.join here rather than self._join |
|
795 | 806 | # because path is arbitrary and user-specified |
|
796 | 807 | files.append(os.path.join(self._rootdir, util.expandpath(path))) |
|
797 | 808 | return files |
|
798 | 809 | |
|
799 | 810 | def _ignorefileandline(self, f): |
|
800 | 811 | files = collections.deque(self._ignorefiles()) |
|
801 | 812 | visited = set() |
|
802 | 813 | while files: |
|
803 | 814 | i = files.popleft() |
|
804 | 815 | patterns = matchmod.readpatternfile( |
|
805 | 816 | i, self._ui.warn, sourceinfo=True |
|
806 | 817 | ) |
|
807 | 818 | for pattern, lineno, line in patterns: |
|
808 | 819 | kind, p = matchmod._patsplit(pattern, b'glob') |
|
809 | 820 | if kind == b"subinclude": |
|
810 | 821 | if p not in visited: |
|
811 | 822 | files.append(p) |
|
812 | 823 | continue |
|
813 | 824 | m = matchmod.match( |
|
814 | 825 | self._root, b'', [], [pattern], warn=self._ui.warn |
|
815 | 826 | ) |
|
816 | 827 | if m(f): |
|
817 | 828 | return (i, lineno, line) |
|
818 | 829 | visited.add(i) |
|
819 | 830 | return (None, -1, b"") |
|
820 | 831 | |
|
821 | 832 | def _walkexplicit(self, match, subrepos): |
|
822 | 833 | """Get stat data about the files explicitly specified by match. |
|
823 | 834 | |
|
824 | 835 | Return a triple (results, dirsfound, dirsnotfound). |
|
825 | 836 | - results is a mapping from filename to stat result. It also contains |
|
826 | 837 | listings mapping subrepos and .hg to None. |
|
827 | 838 | - dirsfound is a list of files found to be directories. |
|
828 | 839 | - dirsnotfound is a list of files that the dirstate thinks are |
|
829 | 840 | directories and that were not found.""" |
|
830 | 841 | |
|
831 | 842 | def badtype(mode): |
|
832 | 843 | kind = _(b'unknown') |
|
833 | 844 | if stat.S_ISCHR(mode): |
|
834 | 845 | kind = _(b'character device') |
|
835 | 846 | elif stat.S_ISBLK(mode): |
|
836 | 847 | kind = _(b'block device') |
|
837 | 848 | elif stat.S_ISFIFO(mode): |
|
838 | 849 | kind = _(b'fifo') |
|
839 | 850 | elif stat.S_ISSOCK(mode): |
|
840 | 851 | kind = _(b'socket') |
|
841 | 852 | elif stat.S_ISDIR(mode): |
|
842 | 853 | kind = _(b'directory') |
|
843 | 854 | return _(b'unsupported file type (type is %s)') % kind |
|
844 | 855 | |
|
845 | 856 | badfn = match.bad |
|
846 | 857 | dmap = self._map |
|
847 | 858 | lstat = os.lstat |
|
848 | 859 | getkind = stat.S_IFMT |
|
849 | 860 | dirkind = stat.S_IFDIR |
|
850 | 861 | regkind = stat.S_IFREG |
|
851 | 862 | lnkkind = stat.S_IFLNK |
|
852 | 863 | join = self._join |
|
853 | 864 | dirsfound = [] |
|
854 | 865 | foundadd = dirsfound.append |
|
855 | 866 | dirsnotfound = [] |
|
856 | 867 | notfoundadd = dirsnotfound.append |
|
857 | 868 | |
|
858 | 869 | if not match.isexact() and self._checkcase: |
|
859 | 870 | normalize = self._normalize |
|
860 | 871 | else: |
|
861 | 872 | normalize = None |
|
862 | 873 | |
|
863 | 874 | files = sorted(match.files()) |
|
864 | 875 | subrepos.sort() |
|
865 | 876 | i, j = 0, 0 |
|
866 | 877 | while i < len(files) and j < len(subrepos): |
|
867 | 878 | subpath = subrepos[j] + b"/" |
|
868 | 879 | if files[i] < subpath: |
|
869 | 880 | i += 1 |
|
870 | 881 | continue |
|
871 | 882 | while i < len(files) and files[i].startswith(subpath): |
|
872 | 883 | del files[i] |
|
873 | 884 | j += 1 |
|
874 | 885 | |
|
875 | 886 | if not files or b'' in files: |
|
876 | 887 | files = [b''] |
|
877 | 888 | # constructing the foldmap is expensive, so don't do it for the |
|
878 | 889 | # common case where files is [''] |
|
879 | 890 | normalize = None |
|
880 | 891 | results = dict.fromkeys(subrepos) |
|
881 | 892 | results[b'.hg'] = None |
|
882 | 893 | |
|
883 | 894 | for ff in files: |
|
884 | 895 | if normalize: |
|
885 | 896 | nf = normalize(ff, False, True) |
|
886 | 897 | else: |
|
887 | 898 | nf = ff |
|
888 | 899 | if nf in results: |
|
889 | 900 | continue |
|
890 | 901 | |
|
891 | 902 | try: |
|
892 | 903 | st = lstat(join(nf)) |
|
893 | 904 | kind = getkind(st.st_mode) |
|
894 | 905 | if kind == dirkind: |
|
895 | 906 | if nf in dmap: |
|
896 | 907 | # file replaced by dir on disk but still in dirstate |
|
897 | 908 | results[nf] = None |
|
898 | 909 | foundadd((nf, ff)) |
|
899 | 910 | elif kind == regkind or kind == lnkkind: |
|
900 | 911 | results[nf] = st |
|
901 | 912 | else: |
|
902 | 913 | badfn(ff, badtype(kind)) |
|
903 | 914 | if nf in dmap: |
|
904 | 915 | results[nf] = None |
|
905 | 916 | except OSError as inst: # nf not found on disk - it is dirstate only |
|
906 | 917 | if nf in dmap: # does it exactly match a missing file? |
|
907 | 918 | results[nf] = None |
|
908 | 919 | else: # does it match a missing directory? |
|
909 | 920 | if self._map.hasdir(nf): |
|
910 | 921 | notfoundadd(nf) |
|
911 | 922 | else: |
|
912 | 923 | badfn(ff, encoding.strtolocal(inst.strerror)) |
|
913 | 924 | |
|
914 | 925 | # match.files() may contain explicitly-specified paths that shouldn't |
|
915 | 926 | # be taken; drop them from the list of files found. dirsfound/notfound |
|
916 | 927 | # aren't filtered here because they will be tested later. |
|
917 | 928 | if match.anypats(): |
|
918 | 929 | for f in list(results): |
|
919 | 930 | if f == b'.hg' or f in subrepos: |
|
920 | 931 | # keep sentinel to disable further out-of-repo walks |
|
921 | 932 | continue |
|
922 | 933 | if not match(f): |
|
923 | 934 | del results[f] |
|
924 | 935 | |
|
925 | 936 | # Case insensitive filesystems cannot rely on lstat() failing to detect |
|
926 | 937 | # a case-only rename. Prune the stat object for any file that does not |
|
927 | 938 | # match the case in the filesystem, if there are multiple files that |
|
928 | 939 | # normalize to the same path. |
|
929 | 940 | if match.isexact() and self._checkcase: |
|
930 | 941 | normed = {} |
|
931 | 942 | |
|
932 | 943 | for f, st in pycompat.iteritems(results): |
|
933 | 944 | if st is None: |
|
934 | 945 | continue |
|
935 | 946 | |
|
936 | 947 | nc = util.normcase(f) |
|
937 | 948 | paths = normed.get(nc) |
|
938 | 949 | |
|
939 | 950 | if paths is None: |
|
940 | 951 | paths = set() |
|
941 | 952 | normed[nc] = paths |
|
942 | 953 | |
|
943 | 954 | paths.add(f) |
|
944 | 955 | |
|
945 | 956 | for norm, paths in pycompat.iteritems(normed): |
|
946 | 957 | if len(paths) > 1: |
|
947 | 958 | for path in paths: |
|
948 | 959 | folded = self._discoverpath( |
|
949 | 960 | path, norm, True, None, self._map.dirfoldmap |
|
950 | 961 | ) |
|
951 | 962 | if path != folded: |
|
952 | 963 | results[path] = None |
|
953 | 964 | |
|
954 | 965 | return results, dirsfound, dirsnotfound |
|
955 | 966 | |
|
956 | 967 | def walk(self, match, subrepos, unknown, ignored, full=True): |
|
957 | 968 | """ |
|
958 | 969 | Walk recursively through the directory tree, finding all files |
|
959 | 970 | matched by match. |
|
960 | 971 | |
|
961 | 972 | If full is False, maybe skip some known-clean files. |
|
962 | 973 | |
|
963 | 974 | Return a dict mapping filename to stat-like object (either |
|
964 | 975 | mercurial.osutil.stat instance or return value of os.stat()). |
|
965 | 976 | |
|
966 | 977 | """ |
|
967 | 978 | # full is a flag that extensions that hook into walk can use -- this |
|
968 | 979 | # implementation doesn't use it at all. This satisfies the contract |
|
969 | 980 | # because we only guarantee a "maybe". |
|
970 | 981 | |
|
971 | 982 | if ignored: |
|
972 | 983 | ignore = util.never |
|
973 | 984 | dirignore = util.never |
|
974 | 985 | elif unknown: |
|
975 | 986 | ignore = self._ignore |
|
976 | 987 | dirignore = self._dirignore |
|
977 | 988 | else: |
|
978 | 989 | # if not unknown and not ignored, drop dir recursion and step 2 |
|
979 | 990 | ignore = util.always |
|
980 | 991 | dirignore = util.always |
|
981 | 992 | |
|
982 | 993 | matchfn = match.matchfn |
|
983 | 994 | matchalways = match.always() |
|
984 | 995 | matchtdir = match.traversedir |
|
985 | 996 | dmap = self._map |
|
986 | 997 | listdir = util.listdir |
|
987 | 998 | lstat = os.lstat |
|
988 | 999 | dirkind = stat.S_IFDIR |
|
989 | 1000 | regkind = stat.S_IFREG |
|
990 | 1001 | lnkkind = stat.S_IFLNK |
|
991 | 1002 | join = self._join |
|
992 | 1003 | |
|
993 | 1004 | exact = skipstep3 = False |
|
994 | 1005 | if match.isexact(): # match.exact |
|
995 | 1006 | exact = True |
|
996 | 1007 | dirignore = util.always # skip step 2 |
|
997 | 1008 | elif match.prefix(): # match.match, no patterns |
|
998 | 1009 | skipstep3 = True |
|
999 | 1010 | |
|
1000 | 1011 | if not exact and self._checkcase: |
|
1001 | 1012 | normalize = self._normalize |
|
1002 | 1013 | normalizefile = self._normalizefile |
|
1003 | 1014 | skipstep3 = False |
|
1004 | 1015 | else: |
|
1005 | 1016 | normalize = self._normalize |
|
1006 | 1017 | normalizefile = None |
|
1007 | 1018 | |
|
1008 | 1019 | # step 1: find all explicit files |
|
1009 | 1020 | results, work, dirsnotfound = self._walkexplicit(match, subrepos) |
|
1010 | 1021 | if matchtdir: |
|
1011 | 1022 | for d in work: |
|
1012 | 1023 | matchtdir(d[0]) |
|
1013 | 1024 | for d in dirsnotfound: |
|
1014 | 1025 | matchtdir(d) |
|
1015 | 1026 | |
|
1016 | 1027 | skipstep3 = skipstep3 and not (work or dirsnotfound) |
|
1017 | 1028 | work = [d for d in work if not dirignore(d[0])] |
|
1018 | 1029 | |
|
1019 | 1030 | # step 2: visit subdirectories |
|
1020 | 1031 | def traverse(work, alreadynormed): |
|
1021 | 1032 | wadd = work.append |
|
1022 | 1033 | while work: |
|
1023 | 1034 | tracing.counter('dirstate.walk work', len(work)) |
|
1024 | 1035 | nd = work.pop() |
|
1025 | 1036 | visitentries = match.visitchildrenset(nd) |
|
1026 | 1037 | if not visitentries: |
|
1027 | 1038 | continue |
|
1028 | 1039 | if visitentries == b'this' or visitentries == b'all': |
|
1029 | 1040 | visitentries = None |
|
1030 | 1041 | skip = None |
|
1031 | 1042 | if nd != b'': |
|
1032 | 1043 | skip = b'.hg' |
|
1033 | 1044 | try: |
|
1034 | 1045 | with tracing.log('dirstate.walk.traverse listdir %s', nd): |
|
1035 | 1046 | entries = listdir(join(nd), stat=True, skip=skip) |
|
1036 | 1047 | except OSError as inst: |
|
1037 | 1048 | if inst.errno in (errno.EACCES, errno.ENOENT): |
|
1038 | 1049 | match.bad( |
|
1039 | 1050 | self.pathto(nd), encoding.strtolocal(inst.strerror) |
|
1040 | 1051 | ) |
|
1041 | 1052 | continue |
|
1042 | 1053 | raise |
|
1043 | 1054 | for f, kind, st in entries: |
|
1044 | 1055 | # Some matchers may return files in the visitentries set, |
|
1045 | 1056 | # instead of 'this', if the matcher explicitly mentions them |
|
1046 | 1057 | # and is not an exactmatcher. This is acceptable; we do not |
|
1047 | 1058 | # make any hard assumptions about file-or-directory below |
|
1048 | 1059 | # based on the presence of `f` in visitentries. If |
|
1049 | 1060 | # visitchildrenset returned a set, we can always skip the |
|
1050 | 1061 | # entries *not* in the set it provided regardless of whether |
|
1051 | 1062 | # they're actually a file or a directory. |
|
1052 | 1063 | if visitentries and f not in visitentries: |
|
1053 | 1064 | continue |
|
1054 | 1065 | if normalizefile: |
|
1055 | 1066 | # even though f might be a directory, we're only |
|
1056 | 1067 | # interested in comparing it to files currently in the |
|
1057 | 1068 | # dmap -- therefore normalizefile is enough |
|
1058 | 1069 | nf = normalizefile( |
|
1059 | 1070 | nd and (nd + b"/" + f) or f, True, True |
|
1060 | 1071 | ) |
|
1061 | 1072 | else: |
|
1062 | 1073 | nf = nd and (nd + b"/" + f) or f |
|
1063 | 1074 | if nf not in results: |
|
1064 | 1075 | if kind == dirkind: |
|
1065 | 1076 | if not ignore(nf): |
|
1066 | 1077 | if matchtdir: |
|
1067 | 1078 | matchtdir(nf) |
|
1068 | 1079 | wadd(nf) |
|
1069 | 1080 | if nf in dmap and (matchalways or matchfn(nf)): |
|
1070 | 1081 | results[nf] = None |
|
1071 | 1082 | elif kind == regkind or kind == lnkkind: |
|
1072 | 1083 | if nf in dmap: |
|
1073 | 1084 | if matchalways or matchfn(nf): |
|
1074 | 1085 | results[nf] = st |
|
1075 | 1086 | elif (matchalways or matchfn(nf)) and not ignore( |
|
1076 | 1087 | nf |
|
1077 | 1088 | ): |
|
1078 | 1089 | # unknown file -- normalize if necessary |
|
1079 | 1090 | if not alreadynormed: |
|
1080 | 1091 | nf = normalize(nf, False, True) |
|
1081 | 1092 | results[nf] = st |
|
1082 | 1093 | elif nf in dmap and (matchalways or matchfn(nf)): |
|
1083 | 1094 | results[nf] = None |
|
1084 | 1095 | |
|
1085 | 1096 | for nd, d in work: |
|
1086 | 1097 | # alreadynormed means that processwork doesn't have to do any |
|
1087 | 1098 | # expensive directory normalization |
|
1088 | 1099 | alreadynormed = not normalize or nd == d |
|
1089 | 1100 | traverse([d], alreadynormed) |
|
1090 | 1101 | |
|
1091 | 1102 | for s in subrepos: |
|
1092 | 1103 | del results[s] |
|
1093 | 1104 | del results[b'.hg'] |
|
1094 | 1105 | |
|
1095 | 1106 | # step 3: visit remaining files from dmap |
|
1096 | 1107 | if not skipstep3 and not exact: |
|
1097 | 1108 | # If a dmap file is not in results yet, it was either |
|
1098 | 1109 | # a) not matching matchfn b) ignored, c) missing, or d) under a |
|
1099 | 1110 | # symlink directory. |
|
1100 | 1111 | if not results and matchalways: |
|
1101 | 1112 | visit = [f for f in dmap] |
|
1102 | 1113 | else: |
|
1103 | 1114 | visit = [f for f in dmap if f not in results and matchfn(f)] |
|
1104 | 1115 | visit.sort() |
|
1105 | 1116 | |
|
1106 | 1117 | if unknown: |
|
1107 | 1118 | # unknown == True means we walked all dirs under the roots |
|
1108 | 1119 | # that wasn't ignored, and everything that matched was stat'ed |
|
1109 | 1120 | # and is already in results. |
|
1110 | 1121 | # The rest must thus be ignored or under a symlink. |
|
1111 | 1122 | audit_path = pathutil.pathauditor(self._root, cached=True) |
|
1112 | 1123 | |
|
1113 | 1124 | for nf in iter(visit): |
|
1114 | 1125 | # If a stat for the same file was already added with a |
|
1115 | 1126 | # different case, don't add one for this, since that would |
|
1116 | 1127 | # make it appear as if the file exists under both names |
|
1117 | 1128 | # on disk. |
|
1118 | 1129 | if ( |
|
1119 | 1130 | normalizefile |
|
1120 | 1131 | and normalizefile(nf, True, True) in results |
|
1121 | 1132 | ): |
|
1122 | 1133 | results[nf] = None |
|
1123 | 1134 | # Report ignored items in the dmap as long as they are not |
|
1124 | 1135 | # under a symlink directory. |
|
1125 | 1136 | elif audit_path.check(nf): |
|
1126 | 1137 | try: |
|
1127 | 1138 | results[nf] = lstat(join(nf)) |
|
1128 | 1139 | # file was just ignored, no links, and exists |
|
1129 | 1140 | except OSError: |
|
1130 | 1141 | # file doesn't exist |
|
1131 | 1142 | results[nf] = None |
|
1132 | 1143 | else: |
|
1133 | 1144 | # It's either missing or under a symlink directory |
|
1134 | 1145 | # which we in this case report as missing |
|
1135 | 1146 | results[nf] = None |
|
1136 | 1147 | else: |
|
1137 | 1148 | # We may not have walked the full directory tree above, |
|
1138 | 1149 | # so stat and check everything we missed. |
|
1139 | 1150 | iv = iter(visit) |
|
1140 | 1151 | for st in util.statfiles([join(i) for i in visit]): |
|
1141 | 1152 | results[next(iv)] = st |
|
1142 | 1153 | return results |
|
1143 | 1154 | |
|
1144 | 1155 | def _rust_status(self, matcher, list_clean, list_ignored, list_unknown): |
|
1145 | 1156 | # Force Rayon (Rust parallelism library) to respect the number of |
|
1146 | 1157 | # workers. This is a temporary workaround until Rust code knows |
|
1147 | 1158 | # how to read the config file. |
|
1148 | 1159 | numcpus = self._ui.configint(b"worker", b"numcpus") |
|
1149 | 1160 | if numcpus is not None: |
|
1150 | 1161 | encoding.environ.setdefault(b'RAYON_NUM_THREADS', b'%d' % numcpus) |
|
1151 | 1162 | |
|
1152 | 1163 | workers_enabled = self._ui.configbool(b"worker", b"enabled", True) |
|
1153 | 1164 | if not workers_enabled: |
|
1154 | 1165 | encoding.environ[b"RAYON_NUM_THREADS"] = b"1" |
|
1155 | 1166 | |
|
1156 | 1167 | ( |
|
1157 | 1168 | lookup, |
|
1158 | 1169 | modified, |
|
1159 | 1170 | added, |
|
1160 | 1171 | removed, |
|
1161 | 1172 | deleted, |
|
1162 | 1173 | clean, |
|
1163 | 1174 | ignored, |
|
1164 | 1175 | unknown, |
|
1165 | 1176 | warnings, |
|
1166 | 1177 | bad, |
|
1167 | 1178 | traversed, |
|
1168 | 1179 | dirty, |
|
1169 | 1180 | ) = rustmod.status( |
|
1170 | 1181 | self._map._rustmap, |
|
1171 | 1182 | matcher, |
|
1172 | 1183 | self._rootdir, |
|
1173 | 1184 | self._ignorefiles(), |
|
1174 | 1185 | self._checkexec, |
|
1175 | 1186 | self._lastnormaltime, |
|
1176 | 1187 | bool(list_clean), |
|
1177 | 1188 | bool(list_ignored), |
|
1178 | 1189 | bool(list_unknown), |
|
1179 | 1190 | bool(matcher.traversedir), |
|
1180 | 1191 | ) |
|
1181 | 1192 | |
|
1182 | 1193 | self._dirty |= dirty |
|
1183 | 1194 | |
|
1184 | 1195 | if matcher.traversedir: |
|
1185 | 1196 | for dir in traversed: |
|
1186 | 1197 | matcher.traversedir(dir) |
|
1187 | 1198 | |
|
1188 | 1199 | if self._ui.warn: |
|
1189 | 1200 | for item in warnings: |
|
1190 | 1201 | if isinstance(item, tuple): |
|
1191 | 1202 | file_path, syntax = item |
|
1192 | 1203 | msg = _(b"%s: ignoring invalid syntax '%s'\n") % ( |
|
1193 | 1204 | file_path, |
|
1194 | 1205 | syntax, |
|
1195 | 1206 | ) |
|
1196 | 1207 | self._ui.warn(msg) |
|
1197 | 1208 | else: |
|
1198 | 1209 | msg = _(b"skipping unreadable pattern file '%s': %s\n") |
|
1199 | 1210 | self._ui.warn( |
|
1200 | 1211 | msg |
|
1201 | 1212 | % ( |
|
1202 | 1213 | pathutil.canonpath( |
|
1203 | 1214 | self._rootdir, self._rootdir, item |
|
1204 | 1215 | ), |
|
1205 | 1216 | b"No such file or directory", |
|
1206 | 1217 | ) |
|
1207 | 1218 | ) |
|
1208 | 1219 | |
|
1209 | 1220 | for (fn, message) in bad: |
|
1210 | 1221 | matcher.bad(fn, encoding.strtolocal(message)) |
|
1211 | 1222 | |
|
1212 | 1223 | status = scmutil.status( |
|
1213 | 1224 | modified=modified, |
|
1214 | 1225 | added=added, |
|
1215 | 1226 | removed=removed, |
|
1216 | 1227 | deleted=deleted, |
|
1217 | 1228 | unknown=unknown, |
|
1218 | 1229 | ignored=ignored, |
|
1219 | 1230 | clean=clean, |
|
1220 | 1231 | ) |
|
1221 | 1232 | return (lookup, status) |
|
1222 | 1233 | |
|
1223 | 1234 | def status(self, match, subrepos, ignored, clean, unknown): |
|
1224 | 1235 | """Determine the status of the working copy relative to the |
|
1225 | 1236 | dirstate and return a pair of (unsure, status), where status is of type |
|
1226 | 1237 | scmutil.status and: |
|
1227 | 1238 | |
|
1228 | 1239 | unsure: |
|
1229 | 1240 | files that might have been modified since the dirstate was |
|
1230 | 1241 | written, but need to be read to be sure (size is the same |
|
1231 | 1242 | but mtime differs) |
|
1232 | 1243 | status.modified: |
|
1233 | 1244 | files that have definitely been modified since the dirstate |
|
1234 | 1245 | was written (different size or mode) |
|
1235 | 1246 | status.clean: |
|
1236 | 1247 | files that have definitely not been modified since the |
|
1237 | 1248 | dirstate was written |
|
1238 | 1249 | """ |
|
1239 | 1250 | listignored, listclean, listunknown = ignored, clean, unknown |
|
1240 | 1251 | lookup, modified, added, unknown, ignored = [], [], [], [], [] |
|
1241 | 1252 | removed, deleted, clean = [], [], [] |
|
1242 | 1253 | |
|
1243 | 1254 | dmap = self._map |
|
1244 | 1255 | dmap.preload() |
|
1245 | 1256 | |
|
1246 | 1257 | use_rust = True |
|
1247 | 1258 | |
|
1248 | 1259 | allowed_matchers = ( |
|
1249 | 1260 | matchmod.alwaysmatcher, |
|
1250 | 1261 | matchmod.exactmatcher, |
|
1251 | 1262 | matchmod.includematcher, |
|
1252 | 1263 | ) |
|
1253 | 1264 | |
|
1254 | 1265 | if rustmod is None: |
|
1255 | 1266 | use_rust = False |
|
1256 | 1267 | elif self._checkcase: |
|
1257 | 1268 | # Case-insensitive filesystems are not handled yet |
|
1258 | 1269 | use_rust = False |
|
1259 | 1270 | elif subrepos: |
|
1260 | 1271 | use_rust = False |
|
1261 | 1272 | elif sparse.enabled: |
|
1262 | 1273 | use_rust = False |
|
1263 | 1274 | elif not isinstance(match, allowed_matchers): |
|
1264 | 1275 | # Some matchers have yet to be implemented |
|
1265 | 1276 | use_rust = False |
|
1266 | 1277 | |
|
1267 | 1278 | if use_rust: |
|
1268 | 1279 | try: |
|
1269 | 1280 | return self._rust_status( |
|
1270 | 1281 | match, listclean, listignored, listunknown |
|
1271 | 1282 | ) |
|
1272 | 1283 | except rustmod.FallbackError: |
|
1273 | 1284 | pass |
|
1274 | 1285 | |
|
1275 | 1286 | def noop(f): |
|
1276 | 1287 | pass |
|
1277 | 1288 | |
|
1278 | 1289 | dcontains = dmap.__contains__ |
|
1279 | 1290 | dget = dmap.__getitem__ |
|
1280 | 1291 | ladd = lookup.append # aka "unsure" |
|
1281 | 1292 | madd = modified.append |
|
1282 | 1293 | aadd = added.append |
|
1283 | 1294 | uadd = unknown.append if listunknown else noop |
|
1284 | 1295 | iadd = ignored.append if listignored else noop |
|
1285 | 1296 | radd = removed.append |
|
1286 | 1297 | dadd = deleted.append |
|
1287 | 1298 | cadd = clean.append if listclean else noop |
|
1288 | 1299 | mexact = match.exact |
|
1289 | 1300 | dirignore = self._dirignore |
|
1290 | 1301 | checkexec = self._checkexec |
|
1291 | 1302 | copymap = self._map.copymap |
|
1292 | 1303 | lastnormaltime = self._lastnormaltime |
|
1293 | 1304 | |
|
1294 | 1305 | # We need to do full walks when either |
|
1295 | 1306 | # - we're listing all clean files, or |
|
1296 | 1307 | # - match.traversedir does something, because match.traversedir should |
|
1297 | 1308 | # be called for every dir in the working dir |
|
1298 | 1309 | full = listclean or match.traversedir is not None |
|
1299 | 1310 | for fn, st in pycompat.iteritems( |
|
1300 | 1311 | self.walk(match, subrepos, listunknown, listignored, full=full) |
|
1301 | 1312 | ): |
|
1302 | 1313 | if not dcontains(fn): |
|
1303 | 1314 | if (listignored or mexact(fn)) and dirignore(fn): |
|
1304 | 1315 | if listignored: |
|
1305 | 1316 | iadd(fn) |
|
1306 | 1317 | else: |
|
1307 | 1318 | uadd(fn) |
|
1308 | 1319 | continue |
|
1309 | 1320 | |
|
1310 | 1321 | # This is equivalent to 'state, mode, size, time = dmap[fn]' but not |
|
1311 | 1322 | # written like that for performance reasons. dmap[fn] is not a |
|
1312 | 1323 | # Python tuple in compiled builds. The CPython UNPACK_SEQUENCE |
|
1313 | 1324 | # opcode has fast paths when the value to be unpacked is a tuple or |
|
1314 | 1325 | # a list, but falls back to creating a full-fledged iterator in |
|
1315 | 1326 | # general. That is much slower than simply accessing and storing the |
|
1316 | 1327 | # tuple members one by one. |
|
1317 | 1328 | t = dget(fn) |
|
1318 |
state = t |
|
|
1329 | state = t.state | |
|
1319 | 1330 | mode = t[1] |
|
1320 | 1331 | size = t[2] |
|
1321 | 1332 | time = t[3] |
|
1322 | 1333 | |
|
1323 | 1334 | if not st and state in b"nma": |
|
1324 | 1335 | dadd(fn) |
|
1325 | 1336 | elif state == b'n': |
|
1326 | 1337 | if ( |
|
1327 | 1338 | size >= 0 |
|
1328 | 1339 | and ( |
|
1329 | 1340 | (size != st.st_size and size != st.st_size & _rangemask) |
|
1330 | 1341 | or ((mode ^ st.st_mode) & 0o100 and checkexec) |
|
1331 | 1342 | ) |
|
1332 | 1343 | or size == FROM_P2 # other parent |
|
1333 | 1344 | or fn in copymap |
|
1334 | 1345 | ): |
|
1335 | 1346 | if stat.S_ISLNK(st.st_mode) and size != st.st_size: |
|
1336 | 1347 | # issue6456: Size returned may be longer due to |
|
1337 | 1348 | # encryption on EXT-4 fscrypt, undecided. |
|
1338 | 1349 | ladd(fn) |
|
1339 | 1350 | else: |
|
1340 | 1351 | madd(fn) |
|
1341 | 1352 | elif ( |
|
1342 | 1353 | time != st[stat.ST_MTIME] |
|
1343 | 1354 | and time != st[stat.ST_MTIME] & _rangemask |
|
1344 | 1355 | ): |
|
1345 | 1356 | ladd(fn) |
|
1346 | 1357 | elif st[stat.ST_MTIME] == lastnormaltime: |
|
1347 | 1358 | # fn may have just been marked as normal and it may have |
|
1348 | 1359 | # changed in the same second without changing its size. |
|
1349 | 1360 | # This can happen if we quickly do multiple commits. |
|
1350 | 1361 | # Force lookup, so we don't miss such a racy file change. |
|
1351 | 1362 | ladd(fn) |
|
1352 | 1363 | elif listclean: |
|
1353 | 1364 | cadd(fn) |
|
1354 | 1365 | elif state == b'm': |
|
1355 | 1366 | madd(fn) |
|
1356 | 1367 | elif state == b'a': |
|
1357 | 1368 | aadd(fn) |
|
1358 | 1369 | elif state == b'r': |
|
1359 | 1370 | radd(fn) |
|
1360 | 1371 | status = scmutil.status( |
|
1361 | 1372 | modified, added, removed, deleted, unknown, ignored, clean |
|
1362 | 1373 | ) |
|
1363 | 1374 | return (lookup, status) |
|
1364 | 1375 | |
|
1365 | 1376 | def matches(self, match): |
|
1366 | 1377 | """ |
|
1367 | 1378 | return files in the dirstate (in whatever state) filtered by match |
|
1368 | 1379 | """ |
|
1369 | 1380 | dmap = self._map |
|
1370 | 1381 | if rustmod is not None: |
|
1371 | 1382 | dmap = self._map._rustmap |
|
1372 | 1383 | |
|
1373 | 1384 | if match.always(): |
|
1374 | 1385 | return dmap.keys() |
|
1375 | 1386 | files = match.files() |
|
1376 | 1387 | if match.isexact(): |
|
1377 | 1388 | # fast path -- filter the other way around, since typically files is |
|
1378 | 1389 | # much smaller than dmap |
|
1379 | 1390 | return [f for f in files if f in dmap] |
|
1380 | 1391 | if match.prefix() and all(fn in dmap for fn in files): |
|
1381 | 1392 | # fast path -- all the values are known to be files, so just return |
|
1382 | 1393 | # that |
|
1383 | 1394 | return list(files) |
|
1384 | 1395 | return [f for f in dmap if match(f)] |
|
1385 | 1396 | |
|
1386 | 1397 | def _actualfilename(self, tr): |
|
1387 | 1398 | if tr: |
|
1388 | 1399 | return self._pendingfilename |
|
1389 | 1400 | else: |
|
1390 | 1401 | return self._filename |
|
1391 | 1402 | |
|
1392 | 1403 | def savebackup(self, tr, backupname): |
|
1393 | 1404 | '''Save current dirstate into backup file''' |
|
1394 | 1405 | filename = self._actualfilename(tr) |
|
1395 | 1406 | assert backupname != filename |
|
1396 | 1407 | |
|
1397 | 1408 | # use '_writedirstate' instead of 'write' to write changes certainly, |
|
1398 | 1409 | # because the latter omits writing out if transaction is running. |
|
1399 | 1410 | # output file will be used to create backup of dirstate at this point. |
|
1400 | 1411 | if self._dirty or not self._opener.exists(filename): |
|
1401 | 1412 | self._writedirstate( |
|
1402 | 1413 | self._opener(filename, b"w", atomictemp=True, checkambig=True) |
|
1403 | 1414 | ) |
|
1404 | 1415 | |
|
1405 | 1416 | if tr: |
|
1406 | 1417 | # ensure that subsequent tr.writepending returns True for |
|
1407 | 1418 | # changes written out above, even if dirstate is never |
|
1408 | 1419 | # changed after this |
|
1409 | 1420 | tr.addfilegenerator( |
|
1410 | 1421 | b'dirstate', |
|
1411 | 1422 | (self._filename,), |
|
1412 | 1423 | self._writedirstate, |
|
1413 | 1424 | location=b'plain', |
|
1414 | 1425 | ) |
|
1415 | 1426 | |
|
1416 | 1427 | # ensure that pending file written above is unlinked at |
|
1417 | 1428 | # failure, even if tr.writepending isn't invoked until the |
|
1418 | 1429 | # end of this transaction |
|
1419 | 1430 | tr.registertmp(filename, location=b'plain') |
|
1420 | 1431 | |
|
1421 | 1432 | self._opener.tryunlink(backupname) |
|
1422 | 1433 | # hardlink backup is okay because _writedirstate is always called |
|
1423 | 1434 | # with an "atomictemp=True" file. |
|
1424 | 1435 | util.copyfile( |
|
1425 | 1436 | self._opener.join(filename), |
|
1426 | 1437 | self._opener.join(backupname), |
|
1427 | 1438 | hardlink=True, |
|
1428 | 1439 | ) |
|
1429 | 1440 | |
|
1430 | 1441 | def restorebackup(self, tr, backupname): |
|
1431 | 1442 | '''Restore dirstate by backup file''' |
|
1432 | 1443 | # this "invalidate()" prevents "wlock.release()" from writing |
|
1433 | 1444 | # changes of dirstate out after restoring from backup file |
|
1434 | 1445 | self.invalidate() |
|
1435 | 1446 | filename = self._actualfilename(tr) |
|
1436 | 1447 | o = self._opener |
|
1437 | 1448 | if util.samefile(o.join(backupname), o.join(filename)): |
|
1438 | 1449 | o.unlink(backupname) |
|
1439 | 1450 | else: |
|
1440 | 1451 | o.rename(backupname, filename, checkambig=True) |
|
1441 | 1452 | |
|
1442 | 1453 | def clearbackup(self, tr, backupname): |
|
1443 | 1454 | '''Clear backup file''' |
|
1444 | 1455 | self._opener.unlink(backupname) |
@@ -1,479 +1,494 b'' | |||
|
1 | 1 | # parsers.py - Python implementation of parsers.c |
|
2 | 2 | # |
|
3 | 3 | # Copyright 2009 Olivia Mackall <olivia@selenic.com> and others |
|
4 | 4 | # |
|
5 | 5 | # This software may be used and distributed according to the terms of the |
|
6 | 6 | # GNU General Public License version 2 or any later version. |
|
7 | 7 | |
|
8 | 8 | from __future__ import absolute_import |
|
9 | 9 | |
|
10 | 10 | import struct |
|
11 | 11 | import zlib |
|
12 | 12 | |
|
13 | 13 | from ..node import ( |
|
14 | 14 | nullrev, |
|
15 | 15 | sha1nodeconstants, |
|
16 | 16 | ) |
|
17 | 17 | from .. import ( |
|
18 | 18 | error, |
|
19 | 19 | pycompat, |
|
20 | 20 | revlogutils, |
|
21 | 21 | util, |
|
22 | 22 | ) |
|
23 | 23 | |
|
24 | 24 | from ..revlogutils import nodemap as nodemaputil |
|
25 | 25 | from ..revlogutils import constants as revlog_constants |
|
26 | 26 | |
|
27 | 27 | stringio = pycompat.bytesio |
|
28 | 28 | |
|
29 | 29 | |
|
30 | 30 | _pack = struct.pack |
|
31 | 31 | _unpack = struct.unpack |
|
32 | 32 | _compress = zlib.compress |
|
33 | 33 | _decompress = zlib.decompress |
|
34 | 34 | |
|
35 | 35 | |
|
36 | 36 | class dirstatetuple(object): |
|
37 | 37 | """represent a dirstate entry |
|
38 | 38 | |
|
39 | 39 | It contains: |
|
40 | 40 | |
|
41 | 41 | - state (one of 'n', 'a', 'r', 'm') |
|
42 | 42 | - mode, |
|
43 | 43 | - size, |
|
44 | 44 | - mtime, |
|
45 | 45 | """ |
|
46 | 46 | |
|
47 | 47 | __slot__ = ('_state', '_mode', '_size', '_mtime') |
|
48 | 48 | |
|
49 | 49 | def __init__(self, state, mode, size, mtime): |
|
50 | 50 | self._state = state |
|
51 | 51 | self._mode = mode |
|
52 | 52 | self._size = size |
|
53 | 53 | self._mtime = mtime |
|
54 | 54 | |
|
55 | 55 | def __getitem__(self, idx): |
|
56 | 56 | if idx == 0 or idx == -4: |
|
57 | 57 | return self._state |
|
58 | 58 | elif idx == 1 or idx == -3: |
|
59 | 59 | return self._mode |
|
60 | 60 | elif idx == 2 or idx == -2: |
|
61 | 61 | return self._size |
|
62 | 62 | elif idx == 3 or idx == -1: |
|
63 | 63 | return self._mtime |
|
64 | 64 | else: |
|
65 | 65 | raise IndexError(idx) |
|
66 | 66 | |
|
67 | @property | |
|
68 | def state(self): | |
|
69 | """ | |
|
70 | States are: | |
|
71 | n normal | |
|
72 | m needs merging | |
|
73 | r marked for removal | |
|
74 | a marked for addition | |
|
75 | ||
|
76 | XXX This "state" is a bit obscure and mostly a direct expression of the | |
|
77 | dirstatev1 format. It would make sense to ultimately deprecate it in | |
|
78 | favor of the more "semantic" attributes. | |
|
79 | """ | |
|
80 | return self._state | |
|
81 | ||
|
67 | 82 | def v1_state(self): |
|
68 | 83 | """return a "state" suitable for v1 serialization""" |
|
69 | 84 | return self._state |
|
70 | 85 | |
|
71 | 86 | def v1_mode(self): |
|
72 | 87 | """return a "mode" suitable for v1 serialization""" |
|
73 | 88 | return self._mode |
|
74 | 89 | |
|
75 | 90 | def v1_size(self): |
|
76 | 91 | """return a "size" suitable for v1 serialization""" |
|
77 | 92 | return self._size |
|
78 | 93 | |
|
79 | 94 | def v1_mtime(self): |
|
80 | 95 | """return a "mtime" suitable for v1 serialization""" |
|
81 | 96 | return self._mtime |
|
82 | 97 | |
|
83 | 98 | |
|
84 | 99 | def gettype(q): |
|
85 | 100 | return int(q & 0xFFFF) |
|
86 | 101 | |
|
87 | 102 | |
|
88 | 103 | class BaseIndexObject(object): |
|
89 | 104 | # Can I be passed to an algorithme implemented in Rust ? |
|
90 | 105 | rust_ext_compat = 0 |
|
91 | 106 | # Format of an index entry according to Python's `struct` language |
|
92 | 107 | index_format = revlog_constants.INDEX_ENTRY_V1 |
|
93 | 108 | # Size of a C unsigned long long int, platform independent |
|
94 | 109 | big_int_size = struct.calcsize(b'>Q') |
|
95 | 110 | # Size of a C long int, platform independent |
|
96 | 111 | int_size = struct.calcsize(b'>i') |
|
97 | 112 | # An empty index entry, used as a default value to be overridden, or nullrev |
|
98 | 113 | null_item = ( |
|
99 | 114 | 0, |
|
100 | 115 | 0, |
|
101 | 116 | 0, |
|
102 | 117 | -1, |
|
103 | 118 | -1, |
|
104 | 119 | -1, |
|
105 | 120 | -1, |
|
106 | 121 | sha1nodeconstants.nullid, |
|
107 | 122 | 0, |
|
108 | 123 | 0, |
|
109 | 124 | revlog_constants.COMP_MODE_INLINE, |
|
110 | 125 | revlog_constants.COMP_MODE_INLINE, |
|
111 | 126 | ) |
|
112 | 127 | |
|
113 | 128 | @util.propertycache |
|
114 | 129 | def entry_size(self): |
|
115 | 130 | return self.index_format.size |
|
116 | 131 | |
|
117 | 132 | @property |
|
118 | 133 | def nodemap(self): |
|
119 | 134 | msg = b"index.nodemap is deprecated, use index.[has_node|rev|get_rev]" |
|
120 | 135 | util.nouideprecwarn(msg, b'5.3', stacklevel=2) |
|
121 | 136 | return self._nodemap |
|
122 | 137 | |
|
123 | 138 | @util.propertycache |
|
124 | 139 | def _nodemap(self): |
|
125 | 140 | nodemap = nodemaputil.NodeMap({sha1nodeconstants.nullid: nullrev}) |
|
126 | 141 | for r in range(0, len(self)): |
|
127 | 142 | n = self[r][7] |
|
128 | 143 | nodemap[n] = r |
|
129 | 144 | return nodemap |
|
130 | 145 | |
|
131 | 146 | def has_node(self, node): |
|
132 | 147 | """return True if the node exist in the index""" |
|
133 | 148 | return node in self._nodemap |
|
134 | 149 | |
|
135 | 150 | def rev(self, node): |
|
136 | 151 | """return a revision for a node |
|
137 | 152 | |
|
138 | 153 | If the node is unknown, raise a RevlogError""" |
|
139 | 154 | return self._nodemap[node] |
|
140 | 155 | |
|
141 | 156 | def get_rev(self, node): |
|
142 | 157 | """return a revision for a node |
|
143 | 158 | |
|
144 | 159 | If the node is unknown, return None""" |
|
145 | 160 | return self._nodemap.get(node) |
|
146 | 161 | |
|
147 | 162 | def _stripnodes(self, start): |
|
148 | 163 | if '_nodemap' in vars(self): |
|
149 | 164 | for r in range(start, len(self)): |
|
150 | 165 | n = self[r][7] |
|
151 | 166 | del self._nodemap[n] |
|
152 | 167 | |
|
153 | 168 | def clearcaches(self): |
|
154 | 169 | self.__dict__.pop('_nodemap', None) |
|
155 | 170 | |
|
156 | 171 | def __len__(self): |
|
157 | 172 | return self._lgt + len(self._extra) |
|
158 | 173 | |
|
159 | 174 | def append(self, tup): |
|
160 | 175 | if '_nodemap' in vars(self): |
|
161 | 176 | self._nodemap[tup[7]] = len(self) |
|
162 | 177 | data = self._pack_entry(len(self), tup) |
|
163 | 178 | self._extra.append(data) |
|
164 | 179 | |
|
165 | 180 | def _pack_entry(self, rev, entry): |
|
166 | 181 | assert entry[8] == 0 |
|
167 | 182 | assert entry[9] == 0 |
|
168 | 183 | return self.index_format.pack(*entry[:8]) |
|
169 | 184 | |
|
170 | 185 | def _check_index(self, i): |
|
171 | 186 | if not isinstance(i, int): |
|
172 | 187 | raise TypeError(b"expecting int indexes") |
|
173 | 188 | if i < 0 or i >= len(self): |
|
174 | 189 | raise IndexError |
|
175 | 190 | |
|
176 | 191 | def __getitem__(self, i): |
|
177 | 192 | if i == -1: |
|
178 | 193 | return self.null_item |
|
179 | 194 | self._check_index(i) |
|
180 | 195 | if i >= self._lgt: |
|
181 | 196 | data = self._extra[i - self._lgt] |
|
182 | 197 | else: |
|
183 | 198 | index = self._calculate_index(i) |
|
184 | 199 | data = self._data[index : index + self.entry_size] |
|
185 | 200 | r = self._unpack_entry(i, data) |
|
186 | 201 | if self._lgt and i == 0: |
|
187 | 202 | offset = revlogutils.offset_type(0, gettype(r[0])) |
|
188 | 203 | r = (offset,) + r[1:] |
|
189 | 204 | return r |
|
190 | 205 | |
|
191 | 206 | def _unpack_entry(self, rev, data): |
|
192 | 207 | r = self.index_format.unpack(data) |
|
193 | 208 | r = r + ( |
|
194 | 209 | 0, |
|
195 | 210 | 0, |
|
196 | 211 | revlog_constants.COMP_MODE_INLINE, |
|
197 | 212 | revlog_constants.COMP_MODE_INLINE, |
|
198 | 213 | ) |
|
199 | 214 | return r |
|
200 | 215 | |
|
201 | 216 | def pack_header(self, header): |
|
202 | 217 | """pack header information as binary""" |
|
203 | 218 | v_fmt = revlog_constants.INDEX_HEADER |
|
204 | 219 | return v_fmt.pack(header) |
|
205 | 220 | |
|
206 | 221 | def entry_binary(self, rev): |
|
207 | 222 | """return the raw binary string representing a revision""" |
|
208 | 223 | entry = self[rev] |
|
209 | 224 | p = revlog_constants.INDEX_ENTRY_V1.pack(*entry[:8]) |
|
210 | 225 | if rev == 0: |
|
211 | 226 | p = p[revlog_constants.INDEX_HEADER.size :] |
|
212 | 227 | return p |
|
213 | 228 | |
|
214 | 229 | |
|
215 | 230 | class IndexObject(BaseIndexObject): |
|
216 | 231 | def __init__(self, data): |
|
217 | 232 | assert len(data) % self.entry_size == 0, ( |
|
218 | 233 | len(data), |
|
219 | 234 | self.entry_size, |
|
220 | 235 | len(data) % self.entry_size, |
|
221 | 236 | ) |
|
222 | 237 | self._data = data |
|
223 | 238 | self._lgt = len(data) // self.entry_size |
|
224 | 239 | self._extra = [] |
|
225 | 240 | |
|
226 | 241 | def _calculate_index(self, i): |
|
227 | 242 | return i * self.entry_size |
|
228 | 243 | |
|
229 | 244 | def __delitem__(self, i): |
|
230 | 245 | if not isinstance(i, slice) or not i.stop == -1 or i.step is not None: |
|
231 | 246 | raise ValueError(b"deleting slices only supports a:-1 with step 1") |
|
232 | 247 | i = i.start |
|
233 | 248 | self._check_index(i) |
|
234 | 249 | self._stripnodes(i) |
|
235 | 250 | if i < self._lgt: |
|
236 | 251 | self._data = self._data[: i * self.entry_size] |
|
237 | 252 | self._lgt = i |
|
238 | 253 | self._extra = [] |
|
239 | 254 | else: |
|
240 | 255 | self._extra = self._extra[: i - self._lgt] |
|
241 | 256 | |
|
242 | 257 | |
|
243 | 258 | class PersistentNodeMapIndexObject(IndexObject): |
|
244 | 259 | """a Debug oriented class to test persistent nodemap |
|
245 | 260 | |
|
246 | 261 | We need a simple python object to test API and higher level behavior. See |
|
247 | 262 | the Rust implementation for more serious usage. This should be used only |
|
248 | 263 | through the dedicated `devel.persistent-nodemap` config. |
|
249 | 264 | """ |
|
250 | 265 | |
|
251 | 266 | def nodemap_data_all(self): |
|
252 | 267 | """Return bytes containing a full serialization of a nodemap |
|
253 | 268 | |
|
254 | 269 | The nodemap should be valid for the full set of revisions in the |
|
255 | 270 | index.""" |
|
256 | 271 | return nodemaputil.persistent_data(self) |
|
257 | 272 | |
|
258 | 273 | def nodemap_data_incremental(self): |
|
259 | 274 | """Return bytes containing a incremental update to persistent nodemap |
|
260 | 275 | |
|
261 | 276 | This containst the data for an append-only update of the data provided |
|
262 | 277 | in the last call to `update_nodemap_data`. |
|
263 | 278 | """ |
|
264 | 279 | if self._nm_root is None: |
|
265 | 280 | return None |
|
266 | 281 | docket = self._nm_docket |
|
267 | 282 | changed, data = nodemaputil.update_persistent_data( |
|
268 | 283 | self, self._nm_root, self._nm_max_idx, self._nm_docket.tip_rev |
|
269 | 284 | ) |
|
270 | 285 | |
|
271 | 286 | self._nm_root = self._nm_max_idx = self._nm_docket = None |
|
272 | 287 | return docket, changed, data |
|
273 | 288 | |
|
274 | 289 | def update_nodemap_data(self, docket, nm_data): |
|
275 | 290 | """provide full block of persisted binary data for a nodemap |
|
276 | 291 | |
|
277 | 292 | The data are expected to come from disk. See `nodemap_data_all` for a |
|
278 | 293 | produceur of such data.""" |
|
279 | 294 | if nm_data is not None: |
|
280 | 295 | self._nm_root, self._nm_max_idx = nodemaputil.parse_data(nm_data) |
|
281 | 296 | if self._nm_root: |
|
282 | 297 | self._nm_docket = docket |
|
283 | 298 | else: |
|
284 | 299 | self._nm_root = self._nm_max_idx = self._nm_docket = None |
|
285 | 300 | |
|
286 | 301 | |
|
287 | 302 | class InlinedIndexObject(BaseIndexObject): |
|
288 | 303 | def __init__(self, data, inline=0): |
|
289 | 304 | self._data = data |
|
290 | 305 | self._lgt = self._inline_scan(None) |
|
291 | 306 | self._inline_scan(self._lgt) |
|
292 | 307 | self._extra = [] |
|
293 | 308 | |
|
294 | 309 | def _inline_scan(self, lgt): |
|
295 | 310 | off = 0 |
|
296 | 311 | if lgt is not None: |
|
297 | 312 | self._offsets = [0] * lgt |
|
298 | 313 | count = 0 |
|
299 | 314 | while off <= len(self._data) - self.entry_size: |
|
300 | 315 | start = off + self.big_int_size |
|
301 | 316 | (s,) = struct.unpack( |
|
302 | 317 | b'>i', |
|
303 | 318 | self._data[start : start + self.int_size], |
|
304 | 319 | ) |
|
305 | 320 | if lgt is not None: |
|
306 | 321 | self._offsets[count] = off |
|
307 | 322 | count += 1 |
|
308 | 323 | off += self.entry_size + s |
|
309 | 324 | if off != len(self._data): |
|
310 | 325 | raise ValueError(b"corrupted data") |
|
311 | 326 | return count |
|
312 | 327 | |
|
313 | 328 | def __delitem__(self, i): |
|
314 | 329 | if not isinstance(i, slice) or not i.stop == -1 or i.step is not None: |
|
315 | 330 | raise ValueError(b"deleting slices only supports a:-1 with step 1") |
|
316 | 331 | i = i.start |
|
317 | 332 | self._check_index(i) |
|
318 | 333 | self._stripnodes(i) |
|
319 | 334 | if i < self._lgt: |
|
320 | 335 | self._offsets = self._offsets[:i] |
|
321 | 336 | self._lgt = i |
|
322 | 337 | self._extra = [] |
|
323 | 338 | else: |
|
324 | 339 | self._extra = self._extra[: i - self._lgt] |
|
325 | 340 | |
|
326 | 341 | def _calculate_index(self, i): |
|
327 | 342 | return self._offsets[i] |
|
328 | 343 | |
|
329 | 344 | |
|
330 | 345 | def parse_index2(data, inline, revlogv2=False): |
|
331 | 346 | if not inline: |
|
332 | 347 | cls = IndexObject2 if revlogv2 else IndexObject |
|
333 | 348 | return cls(data), None |
|
334 | 349 | cls = InlinedIndexObject |
|
335 | 350 | return cls(data, inline), (0, data) |
|
336 | 351 | |
|
337 | 352 | |
|
338 | 353 | def parse_index_cl_v2(data): |
|
339 | 354 | return IndexChangelogV2(data), None |
|
340 | 355 | |
|
341 | 356 | |
|
342 | 357 | class IndexObject2(IndexObject): |
|
343 | 358 | index_format = revlog_constants.INDEX_ENTRY_V2 |
|
344 | 359 | |
|
345 | 360 | def replace_sidedata_info( |
|
346 | 361 | self, |
|
347 | 362 | rev, |
|
348 | 363 | sidedata_offset, |
|
349 | 364 | sidedata_length, |
|
350 | 365 | offset_flags, |
|
351 | 366 | compression_mode, |
|
352 | 367 | ): |
|
353 | 368 | """ |
|
354 | 369 | Replace an existing index entry's sidedata offset and length with new |
|
355 | 370 | ones. |
|
356 | 371 | This cannot be used outside of the context of sidedata rewriting, |
|
357 | 372 | inside the transaction that creates the revision `rev`. |
|
358 | 373 | """ |
|
359 | 374 | if rev < 0: |
|
360 | 375 | raise KeyError |
|
361 | 376 | self._check_index(rev) |
|
362 | 377 | if rev < self._lgt: |
|
363 | 378 | msg = b"cannot rewrite entries outside of this transaction" |
|
364 | 379 | raise KeyError(msg) |
|
365 | 380 | else: |
|
366 | 381 | entry = list(self[rev]) |
|
367 | 382 | entry[0] = offset_flags |
|
368 | 383 | entry[8] = sidedata_offset |
|
369 | 384 | entry[9] = sidedata_length |
|
370 | 385 | entry[11] = compression_mode |
|
371 | 386 | entry = tuple(entry) |
|
372 | 387 | new = self._pack_entry(rev, entry) |
|
373 | 388 | self._extra[rev - self._lgt] = new |
|
374 | 389 | |
|
375 | 390 | def _unpack_entry(self, rev, data): |
|
376 | 391 | data = self.index_format.unpack(data) |
|
377 | 392 | entry = data[:10] |
|
378 | 393 | data_comp = data[10] & 3 |
|
379 | 394 | sidedata_comp = (data[10] & (3 << 2)) >> 2 |
|
380 | 395 | return entry + (data_comp, sidedata_comp) |
|
381 | 396 | |
|
382 | 397 | def _pack_entry(self, rev, entry): |
|
383 | 398 | data = entry[:10] |
|
384 | 399 | data_comp = entry[10] & 3 |
|
385 | 400 | sidedata_comp = (entry[11] & 3) << 2 |
|
386 | 401 | data += (data_comp | sidedata_comp,) |
|
387 | 402 | |
|
388 | 403 | return self.index_format.pack(*data) |
|
389 | 404 | |
|
390 | 405 | def entry_binary(self, rev): |
|
391 | 406 | """return the raw binary string representing a revision""" |
|
392 | 407 | entry = self[rev] |
|
393 | 408 | return self._pack_entry(rev, entry) |
|
394 | 409 | |
|
395 | 410 | def pack_header(self, header): |
|
396 | 411 | """pack header information as binary""" |
|
397 | 412 | msg = 'version header should go in the docket, not the index: %d' |
|
398 | 413 | msg %= header |
|
399 | 414 | raise error.ProgrammingError(msg) |
|
400 | 415 | |
|
401 | 416 | |
|
402 | 417 | class IndexChangelogV2(IndexObject2): |
|
403 | 418 | index_format = revlog_constants.INDEX_ENTRY_CL_V2 |
|
404 | 419 | |
|
405 | 420 | def _unpack_entry(self, rev, data, r=True): |
|
406 | 421 | items = self.index_format.unpack(data) |
|
407 | 422 | entry = items[:3] + (rev, rev) + items[3:8] |
|
408 | 423 | data_comp = items[8] & 3 |
|
409 | 424 | sidedata_comp = (items[8] >> 2) & 3 |
|
410 | 425 | return entry + (data_comp, sidedata_comp) |
|
411 | 426 | |
|
412 | 427 | def _pack_entry(self, rev, entry): |
|
413 | 428 | assert entry[3] == rev, entry[3] |
|
414 | 429 | assert entry[4] == rev, entry[4] |
|
415 | 430 | data = entry[:3] + entry[5:10] |
|
416 | 431 | data_comp = entry[10] & 3 |
|
417 | 432 | sidedata_comp = (entry[11] & 3) << 2 |
|
418 | 433 | data += (data_comp | sidedata_comp,) |
|
419 | 434 | return self.index_format.pack(*data) |
|
420 | 435 | |
|
421 | 436 | |
|
422 | 437 | def parse_index_devel_nodemap(data, inline): |
|
423 | 438 | """like parse_index2, but alway return a PersistentNodeMapIndexObject""" |
|
424 | 439 | return PersistentNodeMapIndexObject(data), None |
|
425 | 440 | |
|
426 | 441 | |
|
427 | 442 | def parse_dirstate(dmap, copymap, st): |
|
428 | 443 | parents = [st[:20], st[20:40]] |
|
429 | 444 | # dereference fields so they will be local in loop |
|
430 | 445 | format = b">cllll" |
|
431 | 446 | e_size = struct.calcsize(format) |
|
432 | 447 | pos1 = 40 |
|
433 | 448 | l = len(st) |
|
434 | 449 | |
|
435 | 450 | # the inner loop |
|
436 | 451 | while pos1 < l: |
|
437 | 452 | pos2 = pos1 + e_size |
|
438 | 453 | e = _unpack(b">cllll", st[pos1:pos2]) # a literal here is faster |
|
439 | 454 | pos1 = pos2 + e[4] |
|
440 | 455 | f = st[pos2:pos1] |
|
441 | 456 | if b'\0' in f: |
|
442 | 457 | f, c = f.split(b'\0') |
|
443 | 458 | copymap[f] = c |
|
444 | 459 | dmap[f] = dirstatetuple(*e[:4]) |
|
445 | 460 | return parents |
|
446 | 461 | |
|
447 | 462 | |
|
448 | 463 | def pack_dirstate(dmap, copymap, pl, now): |
|
449 | 464 | now = int(now) |
|
450 | 465 | cs = stringio() |
|
451 | 466 | write = cs.write |
|
452 | 467 | write(b"".join(pl)) |
|
453 | 468 | for f, e in pycompat.iteritems(dmap): |
|
454 | 469 | if e[0] == b'n' and e[3] == now: |
|
455 | 470 | # The file was last modified "simultaneously" with the current |
|
456 | 471 | # write to dirstate (i.e. within the same second for file- |
|
457 | 472 | # systems with a granularity of 1 sec). This commonly happens |
|
458 | 473 | # for at least a couple of files on 'update'. |
|
459 | 474 | # The user could change the file without changing its size |
|
460 | 475 | # within the same second. Invalidate the file's mtime in |
|
461 | 476 | # dirstate, forcing future 'status' calls to compare the |
|
462 | 477 | # contents of the file if the size is the same. This prevents |
|
463 | 478 | # mistakenly treating such files as clean. |
|
464 | 479 | e = dirstatetuple(e[0], e[1], e[2], -1) |
|
465 | 480 | dmap[f] = e |
|
466 | 481 | |
|
467 | 482 | if f in copymap: |
|
468 | 483 | f = b"%s\0%s" % (f, copymap[f]) |
|
469 | 484 | e = _pack( |
|
470 | 485 | b">cllll", |
|
471 | 486 | e.v1_state(), |
|
472 | 487 | e.v1_mode(), |
|
473 | 488 | e.v1_size(), |
|
474 | 489 | e.v1_mtime(), |
|
475 | 490 | len(f), |
|
476 | 491 | ) |
|
477 | 492 | write(e) |
|
478 | 493 | write(f) |
|
479 | 494 | return cs.getvalue() |
General Comments 0
You need to be logged in to leave comments.
Login now